THE VALIDATION OF THE ANGER IMPLICIT ASSOCIATION TEST

A Dissertation

by

RAFAEL CUELLAR, JR.

Submitted to the Office of Graduate Studies of Texas A&M University

in partial fulfillment of the requirements for the degree of

DOCTOR OF PHILOSOPHY

August 2005

Major Subject: Counseling Psychology

THE VALIDATION OF THE ANGER IMPLICIT ASSOCIATION TEST

A Dissertation

by

RAFAEL CUELLAR, JR.

Submitted to the Office of Graduate Studies of Texas A&M University

in partial fulfillment of the requirements for the degree of

DOCTOR OF PHILOSOPHY

Approved by: Chair of Committee, Collie W. Conoley Committee Members, Dan Brossart Linda Castillo Lloyd Korhonen Head of Department, Michael Benz

August 2005

Major Subject: Counseling Psychology

iii

ABSTRACT

The Validation of the Anger Implicit Association

Test. (August 2005)

Rafael Cuellar, Jr., B.A., The University of Texas at

Austin;

M.S., Texas A&M University-Kingsville

Chair of Advisory Committee: Dr. Collie W. Conoley

The present study investigated the Anger IAT as a

valid measure of anger. In order to answer this question

the relationship between the Anger IAT and traditional

measures of anger, anxiety, and self esteem were examined

for convergent and divergent validity. It was hypothesized

that the Anger IAT measure would be moderately to highly

correlated with the State Trait Anger Expression Inventory-

2 (STAXI-2), correlated less with the State-Trait Anxiety

Inventory (STAI), and correlated least with the Rosenberg

Self Esteem Scale (RSES). Additionally, to demonstrate that

the Anger IAT measure reduces a person’s ability to fake

good, social desirability is hypothesized to have a

moderating effect between the Anger IAT and the STAXI-2.

iv

A total of 60 subjects participated in this

investigation, 42 of which were female and 18 were males.

Furthermore, there were 20 Caucasian, 34 Hispanic, and 6

African American participants.

It was found that the Anger IAT was correlated with

several scales of the STAXI-2. The Anger IAT correlated

less with the STAI and least with the RSES. Furthermore,

it was found that the Anger IAT measure reduced the

participant’s ability to fake good.

v

DEDICATION

I dedicate this dissertation to my darling daughter

Victoria Camille Cuellar.

May God bless you and always bring you joy and peace.

vi

ACKNOWLEDGEMENTS

I would like to express my deepest appreciation to the

following people without whom this endeavor would not have

been accomplished.

I extend my gratitude to Collie Conoley for his

patience and friendship and for his guidance regarding the

computer program used to design the Anger IAT. I also thank

Collie for listening and understanding during a difficult

time in my life.

I thank Dan Brossart and Linda Castillo for their

support, guidance, and humor.

Finally, I would like to thank the most important

people in my life, my family.

I thank my sister Nori and niece Lynn for their words

of encouragement and gentle and not so gentle prodding as

to the completion of the project.

I thank my parents for their unconditional love and

faith without which I could not have accomplished this

goal.

To my precious daughter, Victoria, I thank her for

keeping me real, all the timely hugs, and kisses. I thank

you for being such a wonderful daughter and allowing me to

vii

study and write when you would have rather been at the

pool.

I thank God for placing all of these people in my

path. May God bless all of you.

viii

TABLE OF CONTENTS

Page

ABSTRACT …………………………………………………………………………………………………………………… iii

DEDICATION…………………………………………………………………………………………………………………… v

ACKNOWLEDGEMENTS…………………………………………………………………………………………………… vi

TABLE OF CONTENTS………………………………………………………………………………………………… viii

LIST OF FIGURES……………………………………………………………………………………………………… xi

LIST OF TABLES………………………………………………………………………………………………………… xii

CHAPTER

I INTRODUCTION………………………………………………………………………………… 1

Anger: Concepts and Definitions…………………… 1 Assessment of Anger…………………………………………………… 2 Implicit Measures of Attitudes……………………… 3 Statement of Problem………………………………………………… 8 Purpose of Study…………………………………………………………… 9 Research Question………………………………………………………… 10 Organization of Study……………………………………………… 11

II PROBLEM……………………………………………………………………………………………… 12

Anger Concepts and Definitions……………………… 12 Assessment of Anger…………………………………………………… 14 Anxiety…………………………………………………………………………………… 17 Self-Esteem………………………………………………………………………… 20 Automatic Activation of Attitude………………… 21 Implicit Association Test…………………………………… 28 The Validity of the IAT………………………………………… 31 Attitude Studies…………………………………………………………… 33 Purpose of Study…………………………………………………………… 38 Research Question………………………………………………………… 39

ix

CHAPTER Page

III METHODS……………………………………………………………………………………………… 41

Participants……………………………………………………………………… 41 Measures………………………………………………………………………………… 42 Areas of Concern…………………………………………………………… 58

IV RESULTS……………………………………………………………………………………………… 60

Descriptive Statistics…………………………………………… 60 Reliability Analysis………………………………………………… 60 IAT Data Analysis………………………………………………………… 61 Hypothesized Relationships………………………………… 62 Pearson r Correlations…………………………………………… 64 Controlling Social Desirability…………………… 68 Post-Hoc Test…………………………………………………………………… 74

V DISCUSSION AND CONCLUSION……………………………………………… 75

Purpose of Study…………………………………………………………… 75 Summary of Results……………………………………………………… 77 Post-Hoc Analysis………………………………………………………… 81 Limitations………………………………………………………………………… 82 Future Recommendations…………………………………………… 82 Conclusion…………………………………………………………………………… 84

REFERENCES…………………………………………………………………………………………………………………… 86

APPENDIX A…………………………………………………………………………………………………………………… 97

APPENDIX B…………………………………………………………………………………………………………………… 98

APPENDIX C…………………………………………………………………………………………………………………… 101

APPENDIX D…………………………………………………………………………………………………………………… 103

APPENDIX E…………………………………………………………………………………………………………………… 104

APPENDIX F…………………………………………………………………………………………………………………… 105

APPENDIX G…………………………………………………………………………………………………………………… 106

APPENDIX H…………………………………………………………………………………………………………………… 107

x

Page

APPENDIX I…………………………………………………………………………………………………………………… 109

APPENDIX J…………………………………………………………………………………………………………………… 111

VITA…………………………………………………………………………………………………………………………………… 113

xi

LIST OF FIGURES

FIGURE Page 1 Instructions for the target concept

discrimination task (“other” vs.“self”)…………………………… 44 2 Sample stimuli for the target concept

discrimination step (“other” vs. “self”)………………………… 44 3 Instructions for the attribute discrimination task

(ANGRY words vs. PEACEFUL words)……………………………………………… 45 4 Sample stimuli for the attribute discrimination

task (ANGRY words vs. PEACEFUL words)………………………………… 45 5 Instructions for the target + attribute combined

task (“other” + ANGRY vs. “self” + PEACEFUL)……………… 46 6 Sample stimuli for the target + attribute combined

task (“other” + ANGRY vs. “self” + PEACEFUL)……………… 47 7 Instructions for the reversed target concept

discrimination task (“self” vs. “other”)………………………… 47 8 Sample stimuli for the target concept

discrimination step (“self” vs. “other”)………………………… 48 9 Instructions for the reversed target + attribute

combined task (“self” + PEACEFUL vs. “other” + ANGRY)…………………………………………………………………………………………………………………… 49

10 Sample stimuli for the target + attribute combined

task (“self” + PEACEFUL vs. “other” + ANGRY)……………… 49

xii

LIST OF TABLES

TABLE Page 1 Expected Correlations between the Anger IAT,

State-Trait Anger Expression Inventory-2, State-Trait Anxiety Inventory, and the Rosenberg Self Esteem Scale.……………………………………………………… 63

2 Pearson r Correlations between the Anger IAT,

Anger Problem, Rosenberg Self-Esteem Scale, State Anxiety Scale, and the Trait Anxiety Scale…………………………………………………………………………………………… 65

3 Pearson r Correlations between the Anger IAT and

the State Anger Subscales of the State-Trait Anger Expression Inventory-2…………………………………………………… 66

4 Pearson r Correlations between the Anger IAT and

the Trait Subscales of the State-Trait Anger Expression Inventory-2…………………………………………………………………… 67

5 Person r Correlations among the Anger IAT

Measure and the Anger Expression, Anger Control, and Anger Index Subscales of the State-Trait Anger Expression Inventory-2…………………………………………………… 69

6 Correlations between the Anger IAT, Anger

Problem, Rosenberg Self Esteem Scale, State Anxiety, and Trait Anxiety when Partialling out Marlowe-Crowne Social Desirability Scale……………………………………………………………………………… 70

7 Correlations between the Anger IAT, State Anger,

State Anger-Feelings, State Anger-Verbal, and State Anger-Physical when Partialling out Marlowe-Crowne Social Desirability Scale ………………… 71

8 Correlations between the Anger IAT, Trait Anger,

Trait Anger-Temperament, and Trait Anger-Angry Reaction when Partialling out the Marlowe-Crowne Social Desirability Scale ………………………………………………………… 72

xiii

TABLE Page 9 Correlations between the Anger IAT, Anger

Expression, Anger Control, and Anger Index Subscales of the State-Trait Anger Expression Inventory-2 when Partialling out the Marlowe-Crowne Social Desirability Scale…………………… 73

1

CHAPTER I

INTRODUCTION

Over the last few decades, society has witnessed a

plethora of examples of brutality and aggression amongst

humans (Aronson, 1995). Examples include the Vietnam War,

followed by the bloody civil wars in Central America, the

genocides in Bosnia, and Rwanda, acts of terrorism both at

home and abroad, gang violence in large urban cities, and

the atrocities that occurred in our American schools such

as Columbine. The list continues to grow everyday.

Anger: Concepts and Definitions

To date, a definition of anger has been difficult to

ascertain because it is so closely related and linked with

aggression, hostility, and violence. For example, anger

has often been defined as a subjective experience that

precludes aggression (Averill, 1982).

Spielberger, Jacobs, Russell, and Crane (1983) noted

that although a plethora of research has been conducted on

the negative impact of anger and hostility on physical and

psychological well-being, definitions of these constructs

___________

This dissertation follows the style of The Journal of Social Psychology.

2

have been ambiguous at best. Moreover, according to these

authors anger is a simpler concept than hostility or

aggression and usually refers to an emotional state that

ranges in intensity from irritation and annoyance to fury

and rage. Spielberger, Jacobs, Russell, and Crane (1983)

also made the distinction that anger and hostility refer to

feelings and attitudes; whereas, aggression implies

destructive behavior aimed towards persons or objects.

In conclusion the psychological aspects of anger and

the behavioral manifestations of aggression have been

researched extensively; however, finding one definition for

anger is difficult. For the purposes of this study, anger

will be defined as an attitude noted by thoughts and

feelings that vary from mild irritation to intense fury.

The rationale for this operational definition of anger will

be discussed in detail in a later section of this

manuscript.

Assessment of Anger

Now that anger has been defined the following section

will briefly illustrate a number of past attempts at

assessing anger. According to Spielberger, Jacobs,

Russell, and Crane (1983) the earliest attempts to assess

anger were through projective techniques, behavioral

3

observations, and clinical interviews. Although behavioral

ratings and clinical interviews garnered valuable

information they are highly subjective and great value is

placed on the expertise of the interviewer.

During the 1970’s three other measurements of anger

emerged: 1) Reaction Inventory, 2) Anger Self-Report, and

3) Anger Inventory (Spielberger, Jacobs, Russell, and

Crane, 1983). However, Biaggio, Supplee, and Curtis (1981)

found that the aforementioned instruments were confounded

with hostility. Moreover, Biaggio (1980) concluded that

the validity of the Reaction Inventory, Anger Self-Report,

and Anger Inventory were limited.

Given the problems associated with anger in the world

and the difficulty defining and measuring anger explicitly,

a new method of measuring anger could be of use today. Two

such methods of measuring attitudes implicitly will be

briefly discussed in the following section that may be of

some help with the current dilemma of measuring anger.

Implicit Measures of Attitudes

According to Greenwald and Banaji (1995), measuring

implicit social cognition is difficult because by

definition they are not accessible through introspection.

Therefore, self report measures are not able to adequately

4

measure social cognitions due to the participant’s desire

to “look good” or answer in a socially desirable manner.

Currently, there are two methods that have been forth to

deal with this measurement challenge.

The first method for measuring attitudes and feelings

indirectly was introduced by Higgins and King (1981) and

Fazio, Powell, and Herr (1983) who demonstrated a

successful methodology in accessing attitudes without

direct questioning about the attitudes. In both studies a

priming technique was utilized that was either applicable

or not applicable to the information that would be judged

later. The results of these studies demonstrated that

priming affected the judgments of individuals only when the

primed information was applicable to the material judged.

These results revealed that it is possible to access a

person’s attitude merely by having him or her observe the

attitude object.

The second method for measuring attitudes and feelings

indirectly is through the Implicit Association Test (IAT;

Greenwald, McGhee, & Schwartz, 1998). This was the approach

used in the current study. The IAT was developed as a tool

for assessing implicit attitudes indirectly. According to

Greenwald et al. the IAT requires the participant to

5

respond to four types of words while only using two

response keys. The specific process has been described

later in the study.

The strength of both models is that it allows one to

go beyond the use of self-report instruments and assess

unconscious attitudes. Although both of these methods have

proven to be effective, effect size comparisons between the

priming method and the IAT demonstrated that the IAT method

has twice the priming method’s sensitivity to evaluative

differences (Greenwald et al., 1998). This is important

because it allows for greater accuracy.

Greenwald and Banaji (1995) operationally defined

implicit attitudes as “introspectively unidentifiable (or

inaccurately identified) traces of past experience that

mediate favorable or unfavorable feeling, thought, or

action toward social objects.” They further stated that

implicit attitudes are automatically activated and

therefore similar to cognitive priming procedures developed

for measuring automatic affect or attitude (e.g. Bargh,

Chaiken, Govender, & Pratto, 1992; Fazio, 1993; Greenwald,

Klinger, & Liu, 1989; Perdue, Dovidio, Gurtman, & Tyler,

1990; Perdue & Gurtman, 1990). Moreover, Greenwald and

Banaji (1995) asserted that the IAT is effective in

6

assessing automatic association even when an individual

would prefer not to have this association. Therefore for

the purposes of this study, anger will be operationally

defined as an attitude noted by thoughts ad feelings that

range from mild irritation to intense fury.

According to Greenwald et al. (1998) the IAT measures

the differential association of two target concepts with an

evaluative attribute. In their study, three experiments

were conducted and the overall purpose was to determine the

IAT’s usefulness as a measure of evaluative associations

that underlie implicit attitudes.

The first experiment paired target concepts with

evaluative associations that were expected to be strong

enough to be automatically activated and highly similar

across individuals. The subjects in this experiment

responded to two target concepts: (a) flower names vs.

insect names and (b) musical instrument names vs. weapon

names. These target concepts were paired with pleasant

meaning words and unpleasant meaning words. The

expectation was that the IAT procedure would reveal

superior performance for combinations that were compatible

than the ones that were incompatible. The results

supported the hypothesized relationship.

7

The second experiment attempted to discriminate

differences between Japanese Americans and Korean Americans

in regard to their evaluative associations to Japanese and

Korean ethnic groups. Furthermore, explicit measures were

used to validate the IAT’s results. The hypothesis was

that the Korean subjects would be slower in performing the

Japanese + pleasant combination than the Korean + pleasant

combination. It was also hypothesized that the Japanese

subjects would be slower in performing the Korean +

pleasant combination than the Japanese + pleasant

combination. These patterns were found to be true.

The purpose of the third experiment was to determine

if the IAT could measure an implicit attitude that might

not be found through a self-report measure. The author’s

hypothesized that the white subjects would display an

implicit attitude difference between white and black

categories. The results demonstrated that the white

subjects responded faster to the white + pleasant

combination than the black + pleasant combination. Five

explicit measures were completed by the subjects and

compared with the results of the IAT. It was verified that

the IAT and explicit measures were weakly correlated.

Greenwald et al. (1998) commented that White Americans may

8

not have a negative association to African Americans but

rather it could be that the White Americans are simply not

familiar with African Americans so that there would be no

opinions.

The results of the three experiments conducted by

Greenwald et al. (1998) were consistent with the author’s

hypothesis that the IAT is sensitive to the automatic

evaluative associations. Furthermore, Greenwald, Banaji,

Rudman, Farnham, Nosek and Mellott (2002), stated that the

problem with self-report measures lay with a subject’s

ability to report private thoughts and feelings

inaccurately. Therefore, by utilizing the IAT one can

determine the attitude of subjects without the concern of

the subject skewing responses in a socially desirable

manner.

Statement of Problem

Anger has been assessed through behavioral

observations, clinical interviews, projective techniques,

and a number of explicit measures as noted in the previous

section. None of these techniques assesses anger both

implicitly and quantitatively. The purpose of the following

study is to develop a measure of anger that is capable of

measuring anger implicitly and quantitatively. A

9

quantitative implicit measure of anger would be helpful

because it does not require a participant to provide self-

knowledge about their anger or their willingness to be open

and honest about reporting their anger.

Purpose of Study

The purpose of this study was to develop and

investigate the validity of an implicit association test

(IAT) measure of anger. This purpose is important because

anger is a psychological construct that has significant

implications to nearly every person in everyday life.

Increasing our ability to measure anger has the potential

to open areas of research that are not presently available.

Although many instruments have been developed to measure

anger, there is not a measure of anger currently available

that measures anger implicitly. The ability to measure

anger implicitly is important because it provides an avenue

to the underlying emotions of an individual. Furthermore,

by tapping these underlying or unconscious emotions one can

measure anger without the possibility of an individual

answering in a social desirable manner or “faking good.”

The following research has put forth such a test.

10

Research Question

This study sought to test the validity of an IAT

measure of anger. In order to answer this question the

following study applied Hepner, Kivlighan, and Wampold’s

(1992) method for establishing construct validity.

Heppner, Kivlighan, and Wampold (1992) stated that

construct validity is difficult to determine; however, one

way to establish it is by examining the relationships

between the scores on an instrument and that of another

instrument intended to measure the same construct as well

as other instruments intended to measure different

constructs. Furthermore, the authors stated that a pattern

should emerge with a stronger association between the

instruments that measure related constructs and weaker

associations existing between instruments that measure

different constructs.

Therefore, it would be expected that the Anger IAT

would be correlated with the State-Trait Anger Expression

Inventory-2, little or no correlation with the State-Trait

Anxiety Inventory, and no correlation with the Rosenberg

Self-Esteem Scale. Furthermore, when controlling for social

desirability the correlation between the Anger IAT and the

State Trait Anger Expression Inventory-2 will increase.

11

Organization of Study

The present work is organized into five chapters.

Chapter I provides an introduction to the definition anger,

anger assessment, and a new method to assess anger. Chapter

II provides a comprehensive review of the literature

relevant to the subject of anger, assessment of anger,

anxiety, automatic activation of attitudes, and the

implicit association test. Chapter III describes the

methodology utilized in the present study. Chapter IV

presents the results of the data analysis. Last, Chapter V

discusses the results and their practical implications.

12

CHAPTER II

PROBLEM

Can angry attitudes be more accurately measured with

new technology than the existing measures? More precisely,

this study will attempt to answer the following question.

Is the Anger Implicit Association Test a valid measure of

anger? The reason this is an important question is because

to date there is not a measure of anger using an implicit

and quantitative method. The following study will attempt

to develop a measurement of anger which can address the

shortcomings of past measurements of anger. However,

before describing this study, a review of the literature on

anger, assessment of anger, anxiety, automatic activation

of attitudes, the Implicit Association Test, the validity

of the Implicit Association Test, and attitude studies will

be undertaken. The chapter ends with the purpose of the study

and the hypothesis

Anger: Concepts and Definitions

To date, a definition of anger has been difficult to

ascertain because it is so closely related and linked with

aggression, hostility, and violence. For example, anger

has often been defined as a subjective experience that

13

precludes aggression (Averill, 1982). He also stated that

anger has been defined as a physiological arousal or as an

intervening variable for aggressive acts. Averill (1982)

defined anger as an emotional syndrome whereby its purpose

is to inflict pain or cause harm. According to Averill

(1982), anger has been a topic of considerable study.

Great thinkers of their time have attempted to define anger

including Plato, Aristotle, Seneca, Lactantius, Aquinas,

and Descartes. For example, Aquinas defined anger as “a

judgment by which punishment is inflicted upon sin” (as

cited in Averill, 1982).

Spielberger, Jacobs, Russell, and Crane (1983) noted

that although a plethora of research has been conducted on

the negative impact of anger and hostility on physical and

psychological well-being, definitions of these constructs

have been ambiguous at best. Moreover, according to

Spielberger, Jacobs, Russell, and Crane anger is a simpler

concept than hostility or aggression and usually refers to

an emotional state that ranges in intensity from irritation

and annoyance to fury and rage. Spielberger, Jacobs,

Russell, and Crane also made the distinction that anger and

hostility refer to feelings and attitudes; whereas,

14

aggression implies destructive behavior aimed towards

persons or objects.

In conclusion the psychological aspects of anger and

hostility and the behavioral manifestations of aggression

have been considered extensively; however, as one can see

it is difficult to find one definition for anger. For the

purposes of this study anger will be defined as an attitude

noted by thoughts and feelings that vary from mild

irritation to intense fury. The rationale for this

operational definition of anger will be discussed in detail

in a later section of this manuscript. Therefore, implicit

anger will be operationally defined as an unconscious or

introspectively unidentifiable thought and feeling towards

an object.

Assessment of Anger

The following section will briefly illustrate a number

of past attempts at assessing anger. According to

Spielberger, Jacobs, Russell, and Crane (1983) the earliest

attempts to assess anger were through projective

techniques, behavioral observations, and clinical

interviews. Although behavioral ratings and clinical

interviews garnered valuable information they are highly

15

subjective and great value is placed on the expertise of

the interviewer.

Projective techniques, such as the Rorschach and the

Thematic Apperception Test were widely used to assess anger

during the 1950’s and 1960’s (Spielberger, Jacobs, Russell,

& Crane, 1983). Although the authors acknowledged that

these measures showed promise, they concluded that they are

time consuming, scoring is difficult and subjective,

reliability is low, and there is not a substantial evidence

of their validity.

During the 1970’s three other measurements of anger

emerged: 1) Reaction Inventory, 2) Anger Self-Report, and

3) Anger Inventory (Spielberger, Jacobs, Russell, and

Crane, 1983). However, Biaggio, Supplee, and Curtis (1981)

found that the aforementioned instruments were confounded

with hostility. The correlations between the Reaction

Inventory and the Buss-Durkee Hostility Inventory were r=

.52 and .57. The authors concluded that the validity of

the Reaction Inventory, Anger Self-Report, and Anger

Inventory were limited. Spielberger, Jacobs, Russell, and

Crane (1983) concluded that the aforementioned measures of

anger confounded the experience of anger with situational

determinants of angry reactions. Clearly the consensus on

16

the current science of measuring anger has found the

measures in need of improvement.

In 1971 Evans and Strangeland developed the Reaction

Inventory to assess the amount of anger reaction evoked by

a specific stimulus situation. The inventory consisted of

76 items that subjects were to read and rate on a five

point scale ranging from “not at all” to “very much.”

In 1983 Spielberger discussed the Anger Self-Report.

The author described the Anger Self-Report as a 64-item

questionnaire used to assess angry feelings and the

expression of anger. It is comprised of the following

seven subscales: 1. awareness of anger, 2. general

expression of anger, 3. physical expression of anger, 4.

verbal expression of anger, 5. guilt, 6. condemnation of

anger, and 7. mistrust.

According to Spielberger (1983) the Anger Self-Report

predictive and construct validity has not been firmly

established. Furthermore the author notes that the

inventory has not been frequently used by other

investigators.

In 1975 Novaco developed the Anger Inventory to assess

the extent to which varying situations evoke anger

reactions. The original form consisted of 90 anger

17

provoking statements. The statements were derived during

interviews with college students that focused on events

that had produced angry reactions. After reading each

statement the subject responds to a five point Likert scale

ranging from “not at all” to “very much.”

The State-Trait Anger Expression Inventory-2

(Spielberger, 1996) is a self-report measure of anger that

is the most used measure in the psychology literature. The

State-Trait Anger Expression Inventory-2 views anger as a

complex variable which adds to the richness of the measure.

However, Spielberger’s inventory shares the shortcoming of

the other self-report measures of anger in that a person

must be self-aware of the angry feelings and be willing to

accurately share the awareness. The measure will be

discussed in depth in Chapter 3 of this study. It was one

of the measurements used to determine the validity of the

Implicit Association Test.

Anxiety

This section will briefly discuss the construct of

anxiety. It was included because anxiety will be used to

validate the Implicit Association Test. The measure of

anxiety used in this study will be discussed in detail in

chapter 3.

18

Another construct used in this study was anxiety.

Anxiety was included in the study because it has been a

well-researched construct that is considered distinct and

yet has overlapping properties with anger. Over the years

there have been numerous attempts to clearly conceptualize

anxiety. The following section provides a review of some of

the attempts, which have been undertaken.

Delprato and McGlynn (1984) stated that anxiety had

four common definitions. First, anxiety was described as a

trans-situational personality trait. For example, “Lynn is

anxious.” Second, anxiety may refer to a transient and

situational specific response. An example of this is,

“Lynn is anxious when speaking in public.” Third, anxiety

may be described as an affective experience. For example,

“Lynn feels anxious.” Finally, anxiety may act as a

descriptor of a behavior. For example, “Lynn studied the

material because she was anxious about failing.”

Bellack and Lombardo (1984) observed that anxiety

has been described as a stable characteristic of

personality or is predictable characteristic in a specific

stimulus situation. According to Sarbin (1964), Freud

differentiated between objective and neurotic anxiety.

Objective anxiety was defined as a response to a realistic

19

threat; whereas, neurotic anxiety was defined as an

irrational response to an internal conflict.

Another definition of anxiety was provided by Bellack

and Lombardo (1984) who defined anxiety as a set of

responses involving a combination of cognitive and

physiological reactions. Moreover, the authors stated that

anxiety is elicited by an identifiable stimulus.

According to Spielberger, Gorsuch, Lushene, Vagg, and

Jacobs (1983) the twentieth century has been called the age

of anxiety. Furthermore the authors stated that only since

the 1950’s has anxiety been an area of research. This was

primarily due to the ambiguity of conceptualizing anxiety

and the lack of appropriate instruments to measure anxiety.

However, the authors reported that in the 1950’5 research

on anxiety increased because it was defined theoretically

and a number of scales were created to measure the

construct.

Although there have been a number of definitions of

anxiety, the current study will use the state-trait anxiety

constructs set forth by Spielberger (1966). Spielberger’s

State-Trait Anxiety will be used in this study because it

is currently the most frequently used and cited measurement

of anxiety. Spielberger (1966) stated that anxiety refers

20

to two distinct but related concepts. First, anxiety is

used to describe an unpleasant emotional state

characterized by subjective feelings of tension,

apprehension, nervousness, and worry. Second, anxiety is

used to describe stable individual differences of anxiety

proneness as a personality trait.

As stated previously, anxiety has been correlated with

anger in past studies. Although the correlations were

small it appears that anger and anxiety may have some

unspecified similar properties. In 1983, Spielberger,

Gorsuch, Lushene, Vagg, and Jacobs demonstrated the

associations between the State-Trait Anxiety Inventory and

Personality measures. The association between the State-

Trait Anxiety Inventory and the aggression subscale on

Jackson’s Personality Research Form and the aggression

subscale on the Edwards Personality Preference Schedule

were .34 and .247, respectively.

Self-Esteem

Self-Esteem was also included in the present study.

Self-esteem is a well-known construct that has been

researched extensively. During the 1960’s the fields of

psychiatry, psychology and sociology became interested in

the nature of self-concept (Rosenberg, 1965). Self-esteem

21

is one aspect of self-concept and was defined by Rosenberg

as an individual’s sense of his or her own worth. Since

this is the most broad and widely cited definition of self-

esteem this definition will be used for the current study.

For the purposes of the current study it was hypothesized

that self-esteem would have little to no correlation with

anger which has been the case historically.

Automatic Activation of Attitudes

During the seventies and early eighties, social

psychology had shown an increased interest in the attitude

behavior relationship. Fazio, Powell, and Herr (1983)

stated that the renewed interest was partly due to the

reviews of literature, which questioned whether attitudes

are predictive of future behavior. Furthermore, Fazio et

al. (1983) stated that researchers attempted to identify

moderators of the attitude–behavior relationship that might

clarify when attitudes predict behavior. As a result of

this effort many situational variables (Ajzen & Fishbein,

1973; Schofield, 1975; Warner & De Fleur, 1969,)

personality factors (Zanna, Olsen, & Fazio, 1980) and

attitudinal qualities (Fazio & Zanna, 1978) have been

researched. The next step in the research agenda was the

investigation of the process of how attitude affected

22

behavior (Fazio et al., 1983). Therefore, process research

was designed to shed light on how and why variables affect

the attitude-behavior relationship.

According to Fazio et al. (1983) the first step in

developing a process model of the attitude-behavior

relationship lay in attitude accessibility from memory.

Attitude accessibility was operationally defined as the

ease of the process by which an attitude can be retrieved

from memory upon observation of the stimulus object. The

authors further stated that only when the attitude is

activated and salient can it be influential on the ensuing

behavior. Therefore, Fazio views the concept of attitude

accessibility as the key to understanding the process by

which attitudes guide behavior.

Fazio, Chen, McDonel, and Sherman (1982) suggested

that one possibility of understanding attitude

accessibility could possibly lay in the very definition of

attitude. An attitude is the association between an object

and the evaluation of the object by a person (Fazio, et al,

1982). The object-evaluation association varies in

strength and this strength was found to be a determinant of

the accessibility of the attitude from memory. Therefore,

if there is a strong attitude association with an object,

23

the evaluation of the object will be accessed more easily.

However, if the association is weak it will be more

difficult to access the evaluation.

Two experiments by Fazio et al. (1982) demonstrated

this object-evaluation relationship. The first experiment

examined whether attitudes formed on the basis of a direct

behavioral experience were more accessible than attitudes

formed on the basis of an indirect experience. Twenty-one

subjects were exposed to intellectual puzzles in a direct,

behavioral format versus an indirect, non-behavioral

format. The subjects were told that the experiment would

involve the measurement of their attitudes toward five

intellectual puzzles that would be presented through a

videotape unit. The participants viewed other individuals

working on each type of puzzle. The subjects who were in

the indirect condition were told to view the videotape and

were explicitly told not to attempt to work out the

problems. The subjects in the direct condition were told to

view the videotape as well as work out the puzzles

simultaneously. The subjects then participated in a

response-time task in which they decided whether an

adjective was descriptive of their attitude towards a given

intellectual puzzle. The subjects were induced to

24

repeatedly report their evaluations of the puzzles.

Attitude accessibility was measured through response times

to inquiries about their attitudes towards the intellectual

puzzles. The subjects were presented with a number of

slides followed by an evaluative adjective that the

subjects had to respond to. The subject’s task was to

press a key marked “yes” or “no” depending on whether they

felt the adjective was descriptive of the intellectual

puzzle. The subjects in the direct condition responded

significantly faster than the non-direct condition (Fazio

et al., 1982). In other words, when individuals form an

attitude based on a direct, behavioral task they will react

more quickly than if the experience is indirect and non-

behavioral.

The second experiment by Fazio et al. (1982) employed

seventy-nine subjects who received either an indirect or

direct experience with intellectual puzzles. The subjects

were then asked to rate the interest value of each puzzle

on an eleven point Likert scale. The subjects in the

repeated expression condition completed the attitude

scaling an additional two times. Finally, subjects

participated in a “free-play” exercise that consisted of

three different pages of each type of intellectual puzzle.

25

When the subjects completed this task they filled out an

interest value for each type of puzzle. The results showed

that when the subjects were provided with a “free-play”

opportunity in which they could work with any of the

puzzles they had earlier evaluated, the subjects in the

repeated expression condition were more consistent in their

reported interest and their actual behavior toward puzzle

types than their counterparts in the single-expression

condition. This finding is consistent with what one would

expect in the attitude to behavior process (Fazio et al.,

1983); that is, greater contact with the object allows an

increase in the attitude association

Fazio et al. (1983) reported that “these two

experiments provide us with some confidence concerning the

utility of a conceptual framework that views attitudes as

object-evaluation associations and that emphasizes the

strength of this association as a key determinant of

attitude accessibility.” In summary if there is a strong

association between an object and the ensuing evaluation

then the attitude is accessed more easily.

Fazio et al.’s (1982, 1983) and Fazio and Zanna (1981)

examination of the attitude-behavior relationship furthered

the understanding of how specific variables affect the

26

degree to which an attitude influences a behavior. Fazio

et al. (1982) and Fazio and Zanna (1981) found that

attitudes based upon direct, behavioral experience rather

that indirect, non-behavioral experience displayed greater

consistency between their reported interest and actual

behavior toward puzzle types. Fazio et al. (1982 & 1983)

and Fazio and Zanna (1981) proclaimed that behavior towards

an object is a reflection of a person’s evaluation of that

object. In summary, attitude accessibility has been shown

to serve a key role in how attitudes affect behavior.

Moreover, Fazio demonstrated that if a variable strengthens

the object-evaluation association then the process has an

impact on attitude accessibility as well as on the

attitude-behavior consistency. Fazio et al.’s (1983) next

step was to arrive at a methodology that would allow one to

draw conclusions about attitude accessibility without

directly questioning the individuals about their attitudes.

The studies by Higgins and King (1981) and Fazio et

al. (1983) demonstrated a successful methodology in

accessing attitudes without direct questioning about the

attitudes. In both studies a priming technique was

utilized that was either applicable or not applicable to

the information that would be judged later. The results of

27

these studies demonstrated that priming affected the

judgments of individuals only when the primed information

was applicable to the material judged. These results

revealed that it is possible to access a person’s attitude

merely by having him or her observe the attitude object.

In a related study by Fazio, Jackson, Dunton, and Williams

(1995) a priming technique was utilized to access the

extent to which the presentation of an attitude object

automatically activates an associated evaluation from

memory. The results provided corroboration for Fazio’s

(1990) spontaneous attitude-to-behavior process,

specifically, that evaluations are activated automatically

from memory.

Greenwald, McGhee, & Schwartz (1998) have also

developed an indirect method of assessing judgment

latencies for tasks designed to be facilitated or inhibited

by respondent’s attitudes. They found that attitude

consistent judgments are performed faster than attitude

inconsistent judgments because they are relatively

automatic. This method is important because it does not

depend on a participant’s ability or willingness to report

their attitudes, especially when these attitudes are not

socially acceptable. Rather the automatic activation of an

28

attitude either inhibits or facilitates the respondent’s

ability on a subsequent person perception task, affecting

the speed and accuracy of the respondent’s decision making

(Fazio, 1995).

Implicit Association Test

The Implicit Association Test (IAT; Greenwald, McGhee,

& Schwartz, 1998) was developed as a tool for assessing

implicit attitudes indirectly. The attractiveness of this

model is that it goes beyond the use of self-reports

instruments and assesses unconscious attitudes. Although

both methods have proven to be effective, effect size

comparisons between the priming method and the IAT

demonstrated that the IAT method has twice the priming

method’s sensitivity to evaluative differences (Greenwald

et al., 1998). This is important because it allows for

greater accuracy.

Greenwald and Banaji (1995) operationally defined

implicit attitudes as “introspectively unidentifiable (or

inaccurately identified) traces of past experience that

mediate favorable or unfavorable feeling, thought, or

action toward social objects.” They further stated that

implicit attitudes are automatically activated and

therefore similar to cognitive priming procedures developed

29

for measuring automatic affect or attitude (e.g. Bargh,

Chaiken, Govendi, & Pattro, 1992; Fazio, 1993; Greenwald,

Klinger, & Liu, 1989; Perdue, Dovidio, Gurtman, & Tyler,

1990; Perdue & Gurtman, 1990). Moreover, Greenwald and

Banaji (1995) asserted that the IAT is effective in

assessing automatic association even when an individual

would prefer not to have this association.

According to Greenwald et al. (1998) the IAT measures

the differential association of two target concepts with an

evaluative attribute. In their study, three experiments

were conducted and the overall purpose was to determine the

IAT’s usefulness as a measure of evaluative associations

that underlie implicit attitudes.

The first experiment paired target concepts with

evaluative associations that were expected to be strong

enough to be automatically activated and highly similar

across individuals. The subjects in this experiment

responded to two target concepts: (a) flower names vs.

insect names and (b) musical instrument names vs. weapon

names. These target concepts were paired with pleasant

meaning words and unpleasant meaning words. The

expectation was that the IAT procedure would reveal

superior performance for combinations that were compatible

30

than the ones that were incompatible. The results

supported the hypothesized relationship.

The second experiment attempted to discriminate

differences between Japanese Americans and Korean Americans

in regard to their evaluative associations to Japanese and

Korean ethnic groups. Furthermore, explicit measures were

used to bolster the IAT’s results. The hypothesis was that

the Korean subjects would be slower in performing the

Japanese + pleasant combination than the Korean + pleasant

combination. It was also hypothesized that the Japanese

subjects would be slower in performing the Korean +

pleasant combination than the Japanese + pleasant

combination. These patterns were found to be true.

The purpose of the third experiment was to determine

if the IAT could measure an implicit attitude that might

not be found through a self-report measure. The author’s

hypothesized that the white subjects would display an

implicit attitude difference between white and black

categories. The results demonstrated that the white

subjects responded faster to the white + pleasant

combination than the black + pleasant combination. Five

explicit measures were completed by the subjects and

compared with the results of the IAT. It was verified that

31

the IAT and explicit measures were weakly correlated.

Greenwald et al. (1998) commented that White Americans may

not have a negative association to African Americans but

rather it could be that the White Americans are simply not

familiar with African Americans so that there would be no

opinions.

The results of the three experiments conducted by

Greenwald et al. (1998) were consistent with the author’s

hypothesis that the IAT is sensitive to the automatic

evaluative associations. Furthermore, Greenwald, Banaji,

Rudman, Farnham, Nosek and Mellott (2002), stated that the

problem with self-report measures lay with a subject’s

ability to report private thoughts and feelings

inaccurately. Therefore, by utilizing the IAT one can

determine the attitude of subjects without the concern of

the subject skewing responses in a socially desirable

manner.

The Validity of the IAT

In 2001, Greenwald and Nosek investigated the validity

of the IAT by demonstrating internal validity, convergent

validity, discriminant validity, and predictive validity.

The studies used to demonstrate the validity of the IAT

have all been discussed in previous sections of this

32

chapter. The following section will briefly discuss the

results of these studies. Internal validity was

demonstrated by Greenwald et al. in 1998. The authors

stated that the IAT effect was uninfluenced by whether the

pleasant category was placed on the right hand or the left

hand, by whether the categories contained twenty-five items

or five items, and by whether the response to stimulus

intervals ranged from 150 to 750 milliseconds. The author

concluded that the aforementioned were all indications of

internal validity. One influence that was found to affect

internal validity was the order of administration of the

IAT tasks (i.e., steps 3 and steps 5). The performance of

either of these tasks tends to be faster in step 3 than in

step 5. This procedural effect has been corrected by

counterbalancing the order of these two tasks.

In order to test for convergent validity IATs were

correlated with affective priming procedures introduced by

Fazio, Sanbonmatsu, Powell, and Kardes (1986). Cunningham,

Preacher, and Banaji (2002) reported the relationship

between parallel IAT attitudes and affective priming

measures. The result was a correlation of r = .55 between

the two methods.

33

While investigating discriminant validity, Greenwald

et al. (1998) found correlations between IAT and self

report measure to be weakly positive. A total of 16

correlations demonstrating discriminant validity resulted

in an average correlation of r = .25. Greenwald and Nosek

concluded that there are three possible interpretations as

to why implicit-explicit correlations are reduced: 1. self

reports are inaccurate when subjects are dealing with

politically sensitive criterion, 2. poor introspective

access to attitudes, and 3. homogeneity of attitudes.

Predictive validity has been demonstrated by Greenwald

et al. (1998) when dealing with correlations between group

membership including observations of differences between

Japanese Americans and Korean Americans in implicit

attitudes toward ethnic groups. Greenwald et al. also

demonstrated predictive correlations between IAT measures

with individual differences within groups. For example, the

authors found that ingroup preferences for Japanese and

Korean Americans was predicted by measures of their

immersion in their respective cultures.

Attitude Studies

The following studies are not related to the measure

of anger. However, the following studies are particularly

34

helpful at illustrating how the IAT has been used to

determine association strengths.

Since the development of the IAT, several studies have

used this method of measurement to determine association

strengths between various attitudes. The following section

will provide further details regarding these studies.

In 2000, Greenwald and Farnham conducted an experiment

aiming to evaluate implicit and explicit self-esteem. The

authors used confirmatory factor analysis to test whether

implicit and explicit self-esteem measures converged on a

single construct or identified different constructs all

together. The results of this experiment demonstrated that

the implicit and explicit self-esteem measures had

relatively weak correlations with one another. The authors

inferred that due to these results the implicit and

explicit self-esteem measures were measuring different

constructs.

A second experiment by Greenwald and Farnham (2000)

attempted to validate a measure of implicit gender self

concept. The authors used a known group validation process

to accomplish this goal. The assumption was that the IAT

would produce similar results to prior research using

explicit measures. In other words, Greenwald and Farnham

35

(2000) conducted a study to compare the ability of the IAT

and explicit measures to detect the expected differences

shown in previous research regarding men and women in

masculinity-femininity of self-concept. The results of

this experiment demonstrated that the author’s hypothesis

were correct. Moreover, the effect sizes found were much

larger for the IAT than the explicit measures. This led

the authors to state that the IAT is a better method of

measuring attitudes.

Taken as a whole, these three experiments confirmed

construct validity in three forms: (1) discriminant

validity, (2) predictive validity, and (3) known groups’

validity.

Dasgupta, McGhee, Greenwald, and Banaji (2000) looked

at whether IAT’s findings of pro-white attitudes where

valid due or were due to alternative interpretations such

as greater familiarity with the stimuli. The author’s

hypothesized that participants would associate positive

attributes faster with white than black pictures regardless

of their familiarity with name recognition. The results

demonstrated that participants responded statistically

significantly faster when white pictures were paired with

positive attributes and black pictures were paired with

36

unpleasant attributes than when white pictures were paired

with unpleasant attributes and black pictures were paired

with pleasant attributes. Therefore, the results confirmed

the author’s hypothesis.

In 2002, Hummert, Garstka, O’Brien, Greenwald, and

Mellott conducted a two part study using the Implicit

Association Test to measure age differences in implicit

social cognitions. In their initial study on age

differences they collected self- report measures and IAT

measures on three constructs: (1) age attitude, (2) age

identity, and (3) self esteem from young (18-29), young-old

(55-74), and old-old (75+) participants. The authors

hypothesized that age attitude predicted age differences on

explicit but not implicit measures. They also suggested

that age identity would differ across the self-report

measures and the IAT measures. Finally, the authors stated

that self-esteem measures would remain stable across all

age groups. The results supported their primary hypothesis

that age attitude predicted age differences on explicit but

not implicit measures but rejected their hypothesis

regarding age similarities and differences in implicit

attitudes, identities and self esteem.

37

Hummert, Garstka, O’Brien, Greenwald, and Mellott

(2002) attempted to validate their hypothesis regarding age

related slowing on IAT measures and how z-score

transformations could control for this phenomenon. The

authors conducted an experiment using implicit attitudes

towards insects and flowers that had been used in prior

research (Greenwald et al., 1998). The authors predicted

that all participants, regardless of age differences, would

perceive flowers more favorably than insects. However,

they also concluded that age related slowing would produce

greater response latencies for older than young

participants. The authors used z-score transformations

instead of the usual log transformations to control for

this age related effect. Hummert, Gartska, O’Brien,

Greenwald, and Mellot (2002) found that when using the log

transformation method found that older adults had a

statistically significantly larger IAT effect over younger

adults. However when the z-transformation method was

applied to the data, there were no differences between

older and younger adults on the IAT effect.

Taken together these two studies demonstrated that the

IAT is a useful method in measuring age differences but

38

age-related differences in response latencies must be taken

into account when analyzing the data.

Purpose of Study

Anger has been assessed through behavioral

observations, clinical interviews, projective techniques,

and a number of explicit measures. While these approaches

have value, no extant measures assess anger both implicitly

and quantitatively. The purpose of the following study is

to develop a measure of anger that is capable of measuring

anger implicitly and quantitatively. For the purposes of

this study, implicit anger will be operationally defined as

both a conscious and an unconscious or introspectively

unidentifiable thought and feeling towards an object. The

Anger Implicit Association Test is purported to measure

this construct.

The purpose of this study was to develop and

investigate the validity of an implicit association test

(IAT) measure of anger. This purpose is important because

anger is a psychological construct that has significant

implications to nearly every person in everyday life.

Increasing our ability to measure anger has the potential

to open areas of research that are not presently available.

Although many instruments have been developed to measure

39

anger, there is not a measure of anger currently available

that measures anger implicitly. The ability to measure

anger implicitly is important because it provides an avenue

to the underlying emotions of an individual. Furthermore,

by tapping these underlying or unconscious emotions one can

measure anger without the possibility of an individual

answering in a social desirable manner or “faking good.”

The following research has put forth such a test.

Research Question

This study attempted to answer the following question.

Is the Anger IAT a valid measure of anger? In other words,

the study investigated the possibility that angry attitudes

can be measured by a person perception task.

This study sought to test the validity of an IAT

measure of anger. In order to answer this question the

following study applied Hepner, Kivlighan, and Wampold’s

(1992) method for establishing construct validity.

Heppner, Kivlighan, and Wampold stated that construct

validity is difficult to determine; however, one way to

establish it is by examining the relationships between the

scores on an instrument and that of other instruments

intended to measure the same construct as well as other

instruments intended to measure different constructs.

40

Furthermore, the authors stated that a pattern should

emerge with stronger associations between the instruments

that measure related constructs and weaker associations

existing between instruments that measure different

constructs.

Therefore, it would be expected that the Anger IAT

would be correlated with the State-Trait Anger Expression

Inventory-2, little or no correlation with the State-Trait

Anxiety Inventory, and no correlation with the Rosenberg

Self-Esteem Scale. Furthermore, when controlling for social

desirability the correlation between the Anger IAT and the

State Trait Anger Expression Inventory-2 will increase.

41

CHAPTER III

METHODS

Participants

The participants are drawn from a population of

convenience that is limited to students at a large

southwestern university. This is believed to be

justifiable because these participants should act no

differently than other people on the task. The

participants were the first 60 volunteers from Texas A&M

University and were treated in accordance with the

Institutional Review Board’s standards for the protection

of human subjects.

The following information was ascertained from the

Demographic Data Sheet. The participants ranged in age

from 17 to 58. The mean age of respondents was 28.05 (SD =

11.46). Thirty four of the participants were Hispanic.

Twenty one of the Hispanics were female and thirteen were

male. The mean age of the Hispanic participants was 26.38

(SD = 9.46). There were twenty Caucasian participants.

Sixteen of these participants were female and four were

male. The mean age of the Caucasian participants was 29.65

(SD = 14.83). Six African Americans participated in this

42

study. Five were female and one was male. The mean age of

the African American participants was 32.17 (SD = 8.56).

Measures

Demographic Data Sheet. The Demographic Data Sheet

requested information regarding age, ethnicity, gender, and

self reported anger (see Appendix A).

Anger Implicit Association Test (IAT; Cuellar &

Conoley, unpublished manuscript). The Anger IAT measures

the amount of anger that an individual is currently

feeling. This is measured through reaction times to word

associations that tap unconscious angry attitudes. The

Anger IAT consisted of 40 stimulus words: 10 angry words

(e.g., murder, kill, torture, stab, bloody, bomb, destroy,

shoot, strangulate, terror), and 10 peaceful words (e.g.,

help, smile, wave, kiss, hug, high five, visit, fond, love,

greet, laugh,). The peaceful and angry words employed were

developed by Conoley (personal communication, January

2002).

The Anger IAT was developed by Cuellar and Conoley

(2005). The words used to demonstrate an association

between self and angry or peaceful words was employed by a

study conducted by Conoley (personal communication 2001).

43

These words were then incorporated into the Implicit

Association Test developed by Greenwald et al. (1998).

The Anger IAT was presented to subjects using a

computer so that the reaction times can be accurately

measured. The directions and practice experiences are

presented on the computer and integrated into the

assessment procedure. The process of the Anger IAT

administration can be conceptualized as having five steps.

Step 1. The first step taught the subjects how to

respond to the tasks using two keys. Subjects

distinguished between the target concepts of “self” and

“other” by pressing the right key for “self” and the left

key for “other.” Figure 1 and Figure 2 display the

information that the subjects observed on the computer

screen.

44

The tasks that you will be doing in this experiment involve CATEGORY JUDGMENT. On each trial, a stimulus will be displayed, and you must assign it to one of two categories. You should respond as rapidly as possible in categorizing each stimulus, but don’t respond so fast that you make many errors (occasional errors are okay). The two categories that you are to distinguish are “other” vs. “self” words Press the ‘a’ key if the stimulus is an OTHER word But press the ‘5’ key if the stimulus is a SELF word Figure 1. Instructions for the target concept discrimination task(“other” vs. “self”).

OTHER SELF

Me

Figure 2. Sample stimuli for the target concept discrimination step (“other” vs. “self”).

45

Step 2. The second step introduced the subjects to

the second dimension, the attribute part of the IATA task.

Subjects were asked to differentiate among the stimuli

representing the attribute dimension of angry vs. peaceful.

They were asked to press the right key for angry words and

the left key for peaceful words. Figure 3 and Figure 4

display the info that the subjects observed on the computer

screen.

The two categories that you are to distinguish are: ANGRY vs. PEACEFUL words Press the ‘a’ key if the stimulus is an ANGRY word But Press ‘5’ if the stimulus is a PEACEFUL word

Figure 3. Instructions for the attribute discrimination task (ANGRY words vs. PEACEFUL words).

Angry Peaceful

Kill

Figure 4. Sample stimuli for the attribute discrimination task (ANGRY words vs. PEACEFUL words).

46

Step 3. The third step introduced the subjects to the

combined categorization of the two previous dimensions.

The subjects were asked to categorize items into two

combined categories including the target and attribute

concepts that were previously assigned to the same key in

steps 1 and 2. More specifically, the subjects responded

to the “self” and the peaceful words with the left key and

the “other” and angry words with the right key. Figure 5

and Figure 6 display the information that the subjects

observed on the screen. If a person felt angry this

combination of self with peaceful words would have a

relatively slower reaction time than pairing angry words

with self.

The four categories that you are to distinguish are: OTHER vs. SELF words Or ANGRY vs. PEACEFUL words Press the ‘a’ key if the stimulus is an OTHER word or a ANGRY word. But Press the ‘5’ key if the stimulus is a SELF word or a PEACEFUL word. Figure 5. Instructions for the target + attribute combined task (“other” + ANGRY vs. “self” + PEACEFUL).

47

OTHER SELF

ANGRY PEACEFUL

Love

Figure 6. Sample stimuli for the target + attribute combined task (“other” + Angry vs. “self” + PEACEFUL).

Step 4. The subjects repeated step 1 but with the

responses reversed. This step was done to counter any

effects that may occur due to a subject responding faster

with one finger over the other. Figure 7 and Figure 8

display what the subjects observed on the screen.

The two categories that you are to distinguish are: SELF vs. OTHER Press the ‘a’ key if the stimulus is a PEACEFUL word But press the ‘5’ key if the stimulus is an ANGRY word. Figure 7. Instructions for the reversed target concept discrimination task (SELF vs. OTHER).

48

SELF OTHER

THEM

Figure 8. Sample stimuli for the target concept discrimination step (“self” vs. “other”).

Step 5. The subjects were asked to categorize items

into two combined dimensions that included the target and

attribute concepts that were previously assigned to the

same key. More specifically, the subject responded to the

“self” and Angry words with the left key and the “other”

and the peaceful words with the right key. This step is

similar to step 3 but uses the switched key assignments

used in step 4. Figure 9 and Figure 10 display the

information that the subjects observed on the screen.

49

The four categories that you are to distinguish are: SELF vs. OTHER words Or PEACEFUL vs. ANGRY words Press the ‘a’ key if the stimulus is an SELF word or a PEACEFUL word. But Press the ‘5’ key if the stimulus is a OTHER word or a ANGRY word. Figure 9. Instructions for the reversed target + attribute combined task (“self” + PEACEFUL vs. “other” + ANGRY).

SELF OTHER Angry Peaceful

smile

Figure 10. Sample stimuli for the target + attribute

combined task (“self” + PEACEFUL vs. “other” + ANGRY).

The order described above was given to half of the

subjects. However, step 2 and step 3 were given after step

4 and step 5 for the other half of the subjects. Switching

50

the order of presentation was performed to counterbalance

possible task order effects.

The IAT effect is measured by examining the

differential reaction time between the pairs of words that

fit the subject’s feeling state and the pairs of words that

do not fit the subject’s feeling state. The reaction time

for the pair of words that fit the subject’s state should

be faster. The test blocks used to determine the IAT effect

were blocks 3 and 5. The IAT effect was calculated by

subtracting the mean response latency for performing the

(“OTHER” + PEACEFUL words and “SELF” + ANGRY words) from

“OTHER” + ANGRY words and “SELF” + PEACEFUL words).

Positive scores demonstrate an association between self and

angry words whereas a negative score demonstrates an

association between self and peaceful words.

State-Trait Anger Expression Inventory-2 (STAXI-2;

Spielberger, 1996). The State-Trait Anger Expression

Inventory-2 is a self-report measure which was used to

assess each participant’s state anger and trait anger (see

Appendix B). The state anger scale had three subscales

which measured the intensity of angry feelings a

participant was currently experiencing, the intensity of

current feelings related to the verbal expression of anger,

51

and the intensity of current feelings related to the

physical expression of anger. The trait anger scale had

two subscales that measured the participant’s disposition

to experience anger without specific provocation and the

frequency that angry feelings were experienced in

situations that involved frustration and/or negative

evaluations.

The STAXI-2 may be administered to individuals who are

older than 13. Although there is not a time limit for this

test most subjects usually complete the inventory within 15

minutes (Spielberger, 1988). The STAXI-2 is composed of 57

items which were administered in three parts: (a) state

anger, (b) trait anger and (c) anger expression. State

anger is composed of three subscales totaling 15 items:

Feeling angry, Feel like expressing anger verbally, and

Feel like expressing anger physically. Trait anger is

composed of two subscales, totaling ten items: Angry

temperament and Angry reaction. The anger expression is

composed of two expression constructs. The anger-in

construct related to anger expressed inward toward self and

how often this is experienced. The anger-in construct

consists of eight items. The anger-out construct relates

to anger expressed outwardly towards people and objects in

52

a verbal or physical manner. The anger-out construct

consists of eight items. Anger control relates to two

control constructs. The anger control-out construct

identifies how often a person controls the outward

expression of their angry feelings. The control-out

construct consists of eight items. The anger control-in

construct identifies how often a person controls their

angry feelings by calming down or cooling off. The

control-in construct consists of eight items. Finally,

there is an anger expression index which provides a general

index of anger. The inventory consists of three parts.

The first two parts are composed of 25 items each and

provide scores for the state and trait portions of the

inventory. The last section is composed of 32 items and

provides scores for the expression, control, and index

portions of the inventory.

State anger reliability from the standardization

studies ranged from .87 to .93 and trait anger reliability

ranged from .82 to .84. The reliability ranged for anger

expression-in, anger expression-out, and anger expression-

control from .73 to .86, .73 to .78, and .81 to .85,

respectively (Spielberger, 1988). These ranges of

reliability are considered strong.

53

Validity for the STAXI-2 has been established through

concurrent validity and factor analysis. Trait anger

correlates highly with the Buss-Durkee Hostility Inventory,

.69. Spilberger (1996) conducted a factor analysis on the

STAXI-2. He identified two factors for both male and

females on the state-anger items. The factor loadings

ranged from .34 to .97. The trait-anger items also loaded

on two factors for both male and females. The factor

loadings for trait anger items ranged from .45 to .88 on

factor 1 and from .37 to .88 on factor 2. Next,

Spielberger (1996) identified four factors for the anger

expression and anger control items. The factor loadings

for the anger control-in items ranged from .51 to .92. The

factor loadings for the anger control-out items ranged from

-.23 to .92. The factor loadings for the anger expression-

in items ranged from .37 to .67. Finally, the factor

loadings for the anger expression-out ranged from .18 to

.70.

State-Trait Anxiety Inventory (STAI; Spielberger,

Gorsuch, Lushene, Vagg,& Jacobs, 1970). The State-Trait

Anxiety Inventory assessed the participant’s state anxiety

and trait anxiety (see Appendix C). This inventory is a

self-report measure that usually takes between 10 to 20

54

minutes to administer. It can be administered to people

over grade level 9. The inventory includes separate

measures of state and trait anxiety. State anxiety reflects

a subjective state which may fluctuate over time and can

vary in intensity. In contrast, trait anxiety is a stable

trait and refers to a general tendency to respond in an

anxious manner when a perceived threat is recognized by an

individual.

In the initial standardization studies (Spielberer,

Vagg, Barker, Donham, & Westberry, 1980) reliability of the

inventory was assessed through test-retest intervals

ranging from one hour to 104 days. Trait anxiety scale

coefficients ranged from .65 to .86, whereas the range for

the State anxiety scale was .16 to .62. According to the

author this low level of stability was expected for the

State anxiety scale because it reflects a subjective state

that changes over time and is influenced by situational

factors.

Validity was established through concurrent validity

and factor analysis. In 1970, Spielberger, Gorusch,

Lushene, Vagg and Jacobs reported the following

correlations between the STAI and the Taylor Manifest

Anxiety Scale, the IPAT Anxiety Scale, and the Multiple

55

Affect Adjective Check List respectively as .80, .75, and

.52. In 1980, Spielberer, Vagg, Barker, Donham, &

Westberry, 1980 conducted a factor analysis on the STAI.

After testing 424 tenth grade students, the authors

identified four factors for both females and males. The

factor loadings ranged from .40 to .71.

Rosenberg Self-Esteem Scale (RSES; Rosenberg, 1965).

The Rosenberg Self-Esteem Scale was used to assess the

participant’s self-esteem (see Appendix D). The Rosenberg

Self-Esteem Scale is scored as a Likert scale. The 10 items

that comprised the form were answered on a four point scale

that ranged from strongly agree to strongly disagree. The

scores ranged from 0-30, with 0 indicating the lowest score

possible and 30 indicating the highest score possible. The

larger number indicates lower self-esteem of the subject.

Items 3, 5, 8, 9, and 10 were reverse scored. The original

sample was composed of 5,024 high school juniors and

seniors from 10 randomly selected schools in New York State

(Rosenberg, 1965).

According to Blascovich and Tomaka, (1993) test-retest

correlations are typically in the range of .82 to .88.

Furthermore, in 1987 Rosenberg found that the Cronbach's

alpha for various samples were in the range of .77 to .88.

56

In 1965, Rosenberg stated that although the self-

esteem scale that he created had face validity that was not

enough to establish the adequacy of the scale. However, he

explained that there were no known groups or criterion

groups to validate the scale. Therefore, he defended the

scale on the assumption that that if the scale actually

measures self-esteem then one would expect the scale to be

associated with some other data in a meaningful manner. He

hypothesized that depression accompanies low self-esteem;

therefore, people with low self-esteem should appear more

depressed. He conducted an experiment and found his

hypothesis to be substantiated thereby validating his

scale.

Marlowe-Crowne 2(10) Social Desirability Scale

(MC2(10); Strahan & Gerbasi, 1972). The Marlowe-Crowne

2(10) Social Desirability Scale (1972) assessed the

participant’s tendency to answer in a socially desirable

manner or tendency to present oneself in a good light (see

Appendix E). This scale is composed of 10 items which

respondents answered true or false to behaviors that were

either desirable but uncommon or undesirable but common.

The scores ranged from 0-10 with higher scores representing

a higher need for approval. Alpha coefficients for the MC

57

2(10) for university males, university females, college

females, and British males were .62, .75, .49, and .62,

respectively.

Validity of the scale was determined through cross

validation from the larger version and two shorter

versions, Marlowe-Crown Social Desirability Scale (Crown &

Marlowe, 1960). The cross-validation results obtained

ranges from .80 to .90.

Participants were recruited through educational-

psychology classes and a written flyer requesting

participants for a study regarding word meaning as an

automatic skill (see Appendix F). On arrival, the

participants were told that the experiment involved a

number of tasks that dealt with word meaning as an

automatic skill. They were told that they would be asked

to fill out questionnaires asking about their feelings and

perform a reaction test. When the subjects completed the

informed consent (see Appendix G) they received

computerized instructions that explained the Anger IAT.

The Anger IAT was administered on a Hewlitt-Packard

Pavillion ze4500 desktop. The Anger IAT used the standard

procedure for the Anger IAT. Responses were assigned to

58

the left and right forefinger. The IAT stimuli appeared

vertically and horizontally centered on the screen.

There were a total of five trail blocks employed in

the current study. Each trial block began with

instructions describing the category discriminations and

the assignment of response keys. On each trial the

stimulus item remained on the screen until the subject

responded. After the subject responded the screen would be

blank if the response was correct or the screen would

contain the word “error” if the response was incorrect.

Trials were conducted with a 250 ms interval between

responses to one stimulus and presentation of another.

After the Anger IAT was completed the participants were

administered the explicit measures: Demographic Data

Sheet, STAXI-2, STAI, RSES, and MCSDS. After the materials

were completed the participants had an opportunity to ask

questions or provide remarks regarding the nature of the

test to the experimenter.

Areas of Concern

The majority of the participants when speaking with

the experimenter stated that the Anger IAT was confusing at

first but were able to overcome the initial confusion by

the end of the second trial. More importantly during these

59

feedback sessions, six individuals told the experimenter

that they had an extremely difficult time combining the

“self” and Angry word combinations. However, for the most

part the participants responded quite favorably to the

Anger IAT.

60

CHAPTER IV

RESULTS

Descriptive Statistics

Means and standard deviations were computed for all

variables concerned. Theses are located in Appendix H,

Appendix I, and Appendix J.

Reliability Analysis

Reliability coefficients were calculated for the

measures. Cronbach Alphas for the State-Trait Anxiety

Inventory (Spielberger, 1996) were assessed through inter-

item reliability. The reliability for the following scales

were: 1) state anger, r = .94, 2) state anger-feelings, r =

.92, 3) state anger-verbal, r = .90, 4) state anger-

physical, r = .74, 5) trait anger, r = .85, 6) angry

temperament, r = .90, 7) angry reaction, .71 8) angry

expression-out, r = .76, 9) angry expression-in, .74, 10)

anger control-out, .88, 11) anger control-in, .88, and 12)

anger expression index, r = .75. These are all considered

strong associations. Cronbach alphas were also used to

assess the State-Trait Anxiety Inventory (Spielberger,

Gorsuch, Lushene, Vagg, and Jacobs 1970). The

reliabilities for the state and trait scales were .91 and

.90, respectively. These results are also considered

61

strong inter-item associations. A reliability coefficient

was also used to assess reliability for the Rosenberg Self-

Esteem Scale (Rosenberg, 1965). The Cronbach Alpha for

this scale was .83 which is considered a strong inter-item

correlation. Finally, the Cronbach alpha for the Marlowe-

Crown 2(10) Social desirability scale was .48 which is low

yet consistent with short form of the original Marlow Crown

Social Desirability Scale.

IAT Data Analysis

The data preparation and data analysis followed the

procedure used by Greenwald and Banaji (1995). This was

followed because Greenwald is the leading researcher in the

IAT literature. The first trial of each experimental task

was excluded from analysis because these responses were

typically longer than subsequent trials due to the

participant’s inexperience with the test. Latencies were

then log transformed to reduce the skew. The means were

calculated from the log transformed scores. Finally, 9

subjects were excluded from analysis because of an error

rate higher than 20 percent which is the recommended cut-

off for error rate outlier data (Greenwald & Banaji, 1995).

62

Anger IAT score for each subject was computed by

subtracting the mean response latency for “other” paired

with peaceful words and “self” paired with Angry words

(i.e., the I feel angry stimulus combination) from the mean

response latency for “other” paired with Angry words and

“self” paired with peaceful words (i.e., the I feel

peaceful stimulus combination). Therefore, positive

differences demonstrated stronger associations with anger

whereas negative differences indicated a stronger

association with peace.

This experiment investigated the validity of an IAT

measure of anger. It was assumed that persons with angry

feelings responded faster to pairings of “me” and angry

words than “me” and peaceful words. The mean response

latency for the Anger IAT effect was 427ms (SD = 112).

Hypothesized Relationships

The relationship between Anger IAT, self-report

measure of anger, anxiety, and self-esteem were examined.

It was hypothesized that the Anger IAT would be moderately

to highly correlated with a paper and pencil test of anger,

correlated less with an anxiety measure, and correlated

least with a self-esteem measure.

63

Table 1 depicts the hypothesized relationships among

the Anger IAT, State-Trait Anger Expression Inventory-2,

State-Trait Anxiety Inventory, and the Rosenberg Self

Esteem Scale.

Table 1. Expected Correlations between the Anger IAT, State-Trait Anger Expression Inventory-2, State-Trait Anxiety Inventory, and the Rosenberg Self Esteem Scale. 1 2 3 4 1. IAT -- high low little 2. STAXI-2 -- high moderate 3. STAI -- moderate 4. RSES -- Note. N = 51. IAT = Anger Implicit Association Test: STAXI-2 = State-Trait Anger Expression Inventory-2: RSES = Rosenberg Self Esteem Scale: STAI = State-Trait Anxiety Inventory.

64

Pearson r Correlations

The relationship between Anger IAT and measures of

anger, anxiety, and self-esteem were calculated using

Pearson r correlations. Table 2 presents the correlations

among the Anger IAT, self-reported measure of anger

gathered from the Demographic Data Sheet, Rosenberg Self

Esteem Scale, State Anxiety Scale, and Trait Anxiety Scale.

The results did not demonstrate any statistically

significant correlations between the Anger IAT and the

aforementioned variables. However, there was a positive

correlation between the Rosenberg Self-Esteem Scale (RSES)

and state anxiety scale, r = .42, and the Rosenberg Self-

Esteem Scale (RSES) and trait anxiety, r = .51.

Furthermore, there was a statistically significant

correlation between state and trait anxiety, r = .50.

65

Table 2. Pearson r Correlations between the Anger IAT, Anger Problem, Rosenberg Self Esteem Scale, State Anxiety Scale, and the Trait Anxiety Scale. 1 2 3 4 5 1. IAT -- -.21 -.11 .11 -.20 2. ANGP -- -.17 -.18 -.09 3. RSES -- *.42 *.51 4. SANX -- *.50 5. TANX -- Note. N = 51. IAT = Anger Implicit Association Test: ANGP = Anger Problems: RSES = Rosenberg Self Esteem Scale: SANX = State Anxiety Scale: TANX = Trait Anxiety Scale. * p < .05.

Table 3 provides the Pearson r correlations among the

Anger IAT, and the state anger subscales of the State-Trait

Anger Expression Inventory-2. The results did not

demonstrate any statistically significant correlations

between the Anger IAT and state anger subscales of the

State-Trait Anger Expression Inventory-2 (STAXI-2). The

results did demonstrate statistically significant

66

correlations between the state anger subscales which are to

be expected because they are measuring similar constructs.

Table 3. Pearson r Correlations between the Anger IAT and the State Anger Subscales of the State-Trait Anger Expression Inventory-2. 1 2 3 4 5 1. IAT -- .15 .26 .12 -.06 2. SANG -- *.89 *.97 *.84 3. SANGF -- *.78 *.53 4. SANGV -- *.85 5. SANGP -- Note. N = 51. IAT = Anger Implicit Association Test: SANG = State Anger: SANGF = State Anger-Feelings: SANGV = State Anger-Verbal: SANGP = State Anger-Physical. * p < .05.

Table 4 presents the Pearson r correlations among the

Anger IAT, and the trait anger subscales of the State-Trait

Anger Expression Inventory-2. The results demonstrated

67

that the correlations between the Anger IAT and the trait

anger subscales of the STAXI-2 were not statistically

significant. However, the high correlations between the

trait anger subscales were statistically significant. The

correlations between the subscales were expected because

they were measuring similar constructs.

Table 4. Pearson r Correlations between the Anger IAT and the Trait Subscales of the State-Trait Anger Expression Inventory-2. 1 2 3 4 1. IAT -- -.03 .01 -.05 2. TANG -- *.90 *.81 3. TANGT -- *.58 4. TANGR -- Note. N = 51. IAT = Anger Implicit Association Test: TANG = Trait Anger: TANGT = Trait Anger-Temperament: TANGR = Trait Anger- Angry Reaction. * p < .05.

68

Table 5 provides the Pearson r correlations among the

Anger IAT, anger expression, anger control, and anger index

subscales of the State-Trait Anger Expression Inventory-2.

The correlations between the Anger IAT and the

aforementioned variables were not found to be statistically

significant. However, the correlations between the Anger

Expression-Out (AXO), Anger Expression-In (AXI), and Anger

Expression Index (AXINDEX) were statistically significant

with one another but not with the Anger IAT.

Controlling Social Desirability

The literature suggests that people are often

unwilling to self-report feelings of anger. Therefore, the

69

Table 5. Person r Correlations among the Anger IAT Measure and the Anger Expression, Anger Control, and Anger Index Subscales of the State-Trait Anger Expression Inventory-2. 1 2 3 4 5 6 1. IAT -- .13 .16 .09 .13 .02 2. AXO -- *.62 *-.47 *-.35 *.70 3. AXI -- *-.38 *-.37 *.73 4. ACO -- *.79 *-.86 5. ACI -- *-.82 6. AXINDEX -- Note. N = 51. IAT = Anger Implicit Association Test: AXO = Anger Expression-Out: AXI = Anger Expression-In: ACO = Anger Control-Out: ACI = Anger Control-In: AXINDEX = Anger Expression Index. * p < .05.

analysis was conducted controlling for social desirability

issues. Specifically, the relationship between the Anger

IAT and measures of anger, anxiety, and self-esteem were

examined using partial correlations. The influence of the

Marlowe-Crowne Social Desirability Scale was partialled out

of the relationship between the Anger IAT measure of anger

and the other measures. Table 6 presents the partial

correlations among the Anger IAT, the self-reported measure

of anger gathered from the Demographic Data Sheet, the

70

Rosenberg Self Esteem Scale, the State Anxiety Scale, and

the Trait Anxiety Scale. The correlation between the Anger

IAT and anger problem was statistically significant (r = -

.27) at the .05 level of probability.

Table 6. Correlations between the Anger IAT, Anger Problem, Rosenberg Self Esteem Scale, State Anxiety, and Trait Anxiety when Partialling out Marlowe-Crowne Social Desirability Scale. 1 2 3 4 5 1. IAT -- *-.27 -.04 .19 -.14 2. ANGP -- -.12 -.12 -.02 3. RSES -- *.36 *.46 4. SANX -- *.45 5. TANX -- Note. N = 51. IAT = Anger Implicit Association Test: ANGP = Anger Problems: RSES = Rosenberg Self Esteem Scale: SANX = State Anxiety Scale: TANX = Trait Anxiety Scale. * p < .05.

Table 7 provides the partial correlations among the

Anger IAT, and the state anger subscales of the State-Trait

Anger Expression Inventory-2. The correlations between the

71

Anger IAT and state anger (SANG) (r = .30) and the Anger

IAT and state anger-feeling (SANGF) (r = .41) were

statistically significant at the .05 level of probability.

Table 7. Correlations between the Anger IAT, State Anger, State Anger-Feelings, State Anger-Verbal, and State Anger-Physical when Partialling out Marlowe-Crowne Social Desirability Scale. 1 2 3 4 5 1. IAT -- *.30 *.41 .25 .04 2. SANG -- *.86 *.96 *.80 3. SANGF -- *.72 *.42 4. SANGV -- *.82 5. SANGP -- Note. N = 51. IAT = Anger Implicit Association Test: SANG = State Anger: SANGF = State Anger-Feelings: SANGV = State Anger-Verbal: SANGP = State Anger-Physical. * p < .05.

Table 8 provides the partial correlations among the

Anger IAT, and the trait anger subscales of the State-Trait

Anger Expression Inventory-2. There were no statistically

72

significant correlations between the Anger IAT and the

trait anger subscales of the STAXI-2.

Table 8. Correlations between the Anger IAT, Trait Anger, Trait Anger-Temperament, and Trait Anger-Angry Reaction when Partialling out the Marlowe-Crowne Social Desirability Scale. 1 2 3 4 1. IAT -- .13 .16 .08 2. TANG -- *.85 *.73 3. TANGT -- *.41 4. TANGR -- Note. N = 51. IAT = Anger Implicit Association Test: TANG = Trait Anger: TANGT = Trait Anger-Temperament: TANGR = Trait Anger- Angry Reaction. * p < .05.

Table 9 provides the partial correlations among the

Anger IAT, anger expression, anger control, and anger index

subscales of the State-Trait Anger Expression Inventory-2.

73

There was a statistically significant correlation between

the Anger IAT and the anger expression-out (AXO) (r = .28).

Table 9. Correlations between the Anger IAT, Anger Expression, Anger Control, and Anger Index Subscales of the State-Trait Anger Expression Inventory-2 when Partialling out the Marlowe-Crowne Social Desirability Scale 1 2 3 4 5 6 1. IAT -- *.28 .26 -.04 .03 .19 2. AXO -- *.54 -.26 -.15 *.58 3. AXI -- -.22 -.22 *.68 4. ACO -- *.72 *-.78 5. ACI -- *-.75 6. AXINDEX -- Note. N = 51. IAT = Anger Implicit Association Test: AXO = Anger Expression-Out: AXI = Anger Expression-In: ACO = Anger Control-Out: ACI = Anger Control-In: AXINDEX = Anger Expression Index. * p < .05.

In conclusion, the associated strengths (with effect

size in parenthesis) found between the Anger IAT Effect and

74

the subsequent variables are as follows: 1) anger problem -

.27 (.07), 2) state anger .30 (.09), 3) state anger-

feelings .41 (.17), 4) state anger-verbal .25 (.06), 5)

state anger-physical .04 (.0002), 6) trait anger .13 (.02),

7) trait anger-temperament .16 (.03), 8) trait anger-angry

reaction .08 (.006), 9) anger expression-out .28 (.08), 10)

anger expression-in .26 (.07), 11) anger control out -.04

(.002), 12) and anger control-in .03 (.001), and the 13)

anger expression index .19 (.04).

Post-Hoc Test

Although the following was not part of the original

question, Post-hoc t-tests were applied to all the quasi-

independent variables and a statistically significant

difference was found between response latency and race of

participant. A Post-Hoc T-Test was performed to explore

differences between the Hispanic and Caucasian samples.

There was a statistically significant difference between

Hispanic response latency (M = 213.11, SD = 407.10) and

Caucasian response latency (M = -77.10, SD = 373.35), t(52)

= 2.607, p = .012 (two-tailed) on the Anger IAT. The

Hispanic participants answered faster to associations

(pairings) of “self” and “angry” words than the Caucasian

participants.

75

CHAPTER V

DISCUSSION AND CONCLUSION

The following chapter is organized in four sections.

First, the purpose of the study is summarized. Second, the

results from the study are described and interpreted.

Third, limitations of the study are described. Fourth,

recommendations and implications are presented.

Purpose of Study

The purpose of this study was to develop and

investigate the validity of an implicit association test

(IAT) measure of anger. This purpose is important because

anger is a psychological construct that has significant

implications to nearly every person in everyday life.

Increasing our ability to measure anger has the potential

to open areas of research that are not presently available.

Although many instruments have been developed to measure

anger, there is not a measure of anger currently available

that measures anger implicitly. The ability to measure

anger implicitly is important because it provides an avenue

to the underlying emotions of an individual. Furthermore,

by tapping these emotions one can measure anger without the

possibility of an individual answering in a socially

76

desirable manner or “faking good.” This research has put

forth such a test.

This study sought to test the validity of an IAT

measure of anger. In order to answer this question the

following study applied Hepner, Kivlighan, and Wampold’s

(1992) method for establishing construct validity. Hepner,

Kivlighan, and Wampold stated that construct validity is

difficult to determine; however, one way to establish it is

by examining the relationships between the scores on an

instrument and that of another instrument intended to

measure the same construct as well as other instruments

intended to measure different constructs. By using a

variety of measures a pattern should emerge with a stronger

association between the instruments that measure related

constructs and weaker associations existing between

instruments that measure different constructs.

Therefore, it would be expected that the Anger IAT

would be correlated with the State-Trait Anger Expression

Inventory-2, little or no correlation with the State-Trait

Anxiety Inventory, and no correlation with the Rosenberg

Self-Esteem Scale. Furthermore, when controlling for social

desirability the correlation between the Anger IAT and the

State Trait Anger Expression Inventory-2 will increase.

77

Summary of Results

The Anger IAT was expected to converge with the State-

Trait Anger Expression Inventory-2 (STAXI-2) because both

purport to measure the same construct, anger. Although the

associations between the Anger IAT and the State-Trait

Anger Expression Inventory-2 were not statistically

significant, they were consistent with findings of other

IAT and self-report measures (Greenwald & Farnham, 2000).

In the study by Greenwald and Farnham a correlation of .17

was found between IAT self-esteem measures and an explicit

measure of self-esteem. According to these authors, this

could mean that the two tests are measuring similar but yet

distinct constructs. Furthermore, the average correlation

between IAT and self report measures for 16 items was r =

.25 (Greenwald & Nosek, 2001). Greenwald and Nosek (2001)

concluded that since IAT attitude measures have correlated

weakly with self-report measures then the IAT may assess

constructs that are different from the constructs measured

by the self-report instruments.

The results are consistent with Hepner, Kivlighan, and

Wampold’s (1992) assumptions for construct validity. The

pattern converges with similar constructs (higher

78

correlations) and diverges with differing constructs (lower

correlations).

The Anger IAT was expected to have a low correlation

or no correlation with the State-Trait Anxiety Inventory.

This is important because it helps to demonstrate construct

validity. The assumption is that a low or no correlation

between the Anger IAT and State-Trait Anxiety Inventory

indicates that the Anger IAT is not measuring the same

construct as the State-Trait Anxiety Inventory. The

results of the study demonstrated this assumption and

therefore provide further support of construct validity for

the Anger IAT. The association strengths between the Anger

IAT and state anxiety and trait anxiety were .11 and -.20,

respectively. These results were expected because the Anger

IAT and the State-Trait Anxiety Inventory are measuring two

separate and distinct constructs. Moreover, it was expected

that the Anger IAT and State-Trait Anxiety Inventory would

have a little to no correlation. Past experiments have

demonstrated a low correlation between anger and anxiety

(Jackson, 1967; Zuckerman & Lubin, 1965).

The Anger IAT was expected to have no correlation with

a measure of the Rosenberg Self-Esteem Scale. The results

supported the aforementioned assumption. The association

79

between the Anger IAT and the Rosenberg Self-Esteem Scale

were -.04. The result was expected because the Anger IAT

and the Rosenberg Self-Esteem Scale are measuring different

constructs.

Finally, the correlation between the Anger IAT and the

State-Trait Anger Expression Inventory-2, when controlling

for social desirability, was expected to be larger than

when not controlling for social desirability. The size of

association between the Anger IAT and the State-Trait Anger

Expression Inventory were greater when the Marlowe-Crowne

Social desirability Scale was used as a moderating variable

between the self-report of anger (STAXI-2) and the Anger

IAT. Moreover, the association strengths between the Anger

IAT and the State-Trait Anger Expression Inventory-2 were

statistically significant for three subscales of the State-

Trait Anger Expression Inventory-2. The association

strengths between the Anger IAT and these three variables;

anger expression-out (AXO), state anger (SANG), and state

anger-feelings (SANGF) were .28, .30, and .41,

respectively. The association strengths between the Anger

IAT and the anger expression-out (AXO), state anger (SANG),

and state anger-feelings (SANGF) prior to using the

Marlowe-Crowne Social Desirability scale as a moderating

80

variable were .13, .15, and .26, respectively. Therefore,

the hypothesis that controlling for social desirability

would increase the correlation between the Anger IAT and

the State-Trait Anger Expression Inventory-2 was supported

by the results. The subscales will be discussed in detail

to better identify the nature of the correlations and what

the Anger IAT is measuring.

According to Spielberger (1999) the state anger scale

measures the intensity of angry feelings and how often a

person feels like expressing their anger at a specific

time. The association between the Anger IAT and state

anger subscale demonstrated a low positive relationship.

This finding helps support the hypothesis for convergent

validity but the correlation was lower than expected.

The state anger-feelings scale measures the intensity

of angry feelings that a person is currently experiencing

(Spielberger, 1999). This scale is relatively transient in

nature and ranges from mildly annoyed to furious. The

association between the Anger IAT and state anger-feelings

subscale demonstrated a low positive relationship. Once

again, this result supports the hypothesis for convergent

validity but the correlation was lower than expected.

81

The anger expression-outward scale (AXO) measures how

often angry feelings are expressed either verbally or

physically (Spielberger, 1999). Furthermore, individuals

with high scores on this scale may express their anger by

slamming doors or physically assaulting someone. Verbally,

they may express their anger through insults, criticisms,

profanity and sarcasm. The association between the Anger

IAT and anger expression-out subscale demonstrated a low

positive relationship. This was also lower than expected.

To reiterate: Is the Anger IAT a valid measure of

anger? The results demonstrated that the Anger IAT has a

stronger association with the State-Trait Anger Expression

Inventory-2, less with the State-Trait Anxiety Inventory,

and least with the Rosenberg Self Esteem Scale. Therefore,

the results fit the assumptions set forth for construct

validity by Hepner, Kivlighan, and Wampold (1992).

Furthermore, the results demonstrated that the Anger IAT

may be measuring current or state anger feelings that a

person is experiencing and wants to express verbally or

physically.

Post-Hoc Analysis

Although the following was not part of the original

question, Post-hoc t-tests were applied to all the quasi-

82

independent variables and a statistically significant

difference was found between response latency and race of

participant. Due to the aforementioned results, it appears

that Hispanics reported more intensity of angry feelings

that they were experiencing. Future research may which to

investigate this phenomenon.

Limitations

Ideally, the best method to establish construct

validity is through a multitrait-multimethod matrix design.

This author used a multitrait design to establish construct

validity of the Anger IAT. Given unlimited resources and

time the author would have conducted a multitrait-

multimethod design. However, other studies have used the

multitrait design and the results were valid so this author

felt confident with the multitrait design to establish

construct validity. A further limitation was sample size.

A larger sample size would have allowed for the examination

of anger validity for different ethnic groups, age

differences, and gender differences.

Future Recommendations

Past research on anger has largely depended on self-

report measures (explicit tests). Future research using

83

the Anger Implicit Association Test may provide new

information about anger. The ability to measure anger

without depending on the participant’s willingness to be

open and honest about reporting their attitudes and

feelings could open new understanding to further our

treatment and prevention of anger and aggressive problems.

Further research on the Anger IAT may focus on studies

between known groups of angry individual, such as

individuals who have committed violent crime as compared

with a nonviolent group. This could help demonstrate the

sensitivity of the Anger IAT as a measure that can discern

the differences between various groups and eventually

provide baselines or norms for the different groups.

Future research could also lead to a better understanding

of limits of the Anger IAT measure. It would be

significant to find that the Anger IAT predicted aggressive

acts so that prevention could be focused. Knowing a person

needs help containing heightened amounts of anger could

avert many tragedies. The strength of the Anger IAT may be

that the person need not have insight into his or her

anger. The Anger IAT may be able to help the angry

individual who does not know that he or she is dangerously

angry.

84

Finally, due to the differences in response time found

regarding ethnicity of race, future research could focus on

norming of races for the Anger IAT. A study that compares

the differences between response latency and race could

provide a different baseline and range of scores between

the differing groups being analyzed. Because anger is

correlated with many health concerns ethnic differences

could also signal health concerns.

Conclusion

In conclusion, the construction and development of a

new inventory, Anger Implicit Association Test, was

described in detail. Furthermore, the validity of the

Anger Implicit Association Test was demonstrated through a

multitrait design which tested for construct validity. The

Anger IAT was demonstrated to be a valid measure of anger

that bypasses the censorship of an individual’s desire to

be socially appropriate. The purpose of developing a

measure of anger implicitly is important because it

provides an avenue to gauge an individual’s thoughts and

feelings when the material is socially sensitive. The

ability to tap these underlying or non-conscious cognitions

and emotions implicitly can help measure anger without the

85

possibility of an individual answering in a social

desirable manner or “faking good.”

The anger IAT has several advantages over the State-

Trait Anger Expression Inventory-2. First and foremost,

the Anger IAT removes the issues of social desirability

when measuring anger. Secondly the anger IAT is quicker to

administer and scores automatically. The State-Trait Anger

Expression Inventory takes approximately twice as long to

administer and score. Thirdly, the Anger IAT is free from

scoring error because it scores automatically. Finally, it

is much more difficult to answer falsely on the Anger IAT.

Finally, it appears that the Anger IAT may be

measuring current or state anger. More specifically, the

results demonstrate that the Anger IAT may be measuring

angry thoughts and feelings that a person is experiencing

and perhaps wants to express verbally or physically.

86

REFERENCES

Ajzen, I., & Fishbein, M. (1973). Attitudinal and normative

variables as predictors of specific behaviors. Journal

of Personality and Social Psychology, 27, 41-57.

Aronson, E. (1995). The social animal (7th ed.). New York:

W.H. Freeman and Company.

Averill, J. R. (1982). Anger and aggression an essay on

emotion. New York: Springer-Verlag.

Bargh, J.A., Chaiken, S., Govender, R., & Pratto, F.

(1992). The generality of the automatic attitude

activation effect. Journal of Personality and Social

Psychology, 62, 893-912.

Bellack, A.S., & Lombardo, T.W. (1984). Behavorial theories

and treatment of anxiety. New York. US: Plenum Press.

Biaggio, M.K. (1980). Assessment of anger arousal. Journal

of Personality Assessment, 44, 289-298.

Biaggio, M.K., Supplee, K., & Curtis, N. (1981).

Reliability and validity of four anger scales. Journal

of Personality Assessment, 45(6), 639-648.

87

Blascovich, J., & Tomaka, J. (1993). Measures of self

esteem. In J. P. Robinson, P. R. Shaver, & L. S.

Wrightsman (Eds.), Measures of personality and social

psychological attitudes (3rd ed., pp. 115-160). Ann

Arbor, MI: Institute for Social Research. Chapter

Citation

C. Conoley (personal communication, January 2002)

C. Conoley (personal communication, 2001)

Cuellar, R., & Conoley, C. (2005). Anger implicit

association test. Unpublished manuscript.

Cunningham, W.A., Preacher, K.J., Banaji, M.R. (2002).

Implicit attitude measures: Consistency, stability,

and convergent validity. Psychological Science.

Dasgupta, N., McGhee, D. E., Greenwald, A. G., & Banaji,

M.R. (2000). Automatic preference for white americans:

Eliminating the familiarity explanation. Journal of

Experimental Social Psychology, 36, 316-328.

Delprato, D. J., & McGlynn, F. D. (1984). Behavioral

theories of anxiety disorders. In S. M. Turner (Ed.),

88

Behavioral treatment of anxiety disorders. New York:

Plenum.

Evans, D.R., & Strangeland, M. (1971). Development of the

reaction inventory to measure anger. Psychological

Reports, 29, 412-414.

Fazio, R.H. (1995). Attitudes as object-evaluation

associations: Determinants, consequences, and

correlates of attitude accessibility. In R.E. Petty &

J.A. Krosnick (Eds.), Attitude strength:

Antecedents and consequences (pp. 247-282).

Hillsdale, NJ: Erlbaum.

Fazio, R.H. (1993). Variability in the likelihood of

automatic attitude activation: Data re-analysis and

commentary on Bargh, Chaiken, Govender, and Pratto

(1992). Journal of Personality and Social Psychology,

64, 753-758, 764-765.

Fazio, R.H. (1990). Multiple processes by which attitudes

guide behavior: The mode model as an integrative

framework. In M.P. Zanna (Ed.), Advances in

experimental social psychology. (Vol. 23, pp. 75-109).

New York: Academic Press.

89

Fazio, R. H., Chen, J. M., McDonel, E. C., & Sherman, S. J.

(1982). Attitude accessibility, attitude-behavior

consistency, and the strength of the object evaluation

association. Journal of Experimental Social

Psychology, 18, 339-357.

Fazio, R. H., Jackson, J. R. Dunton, B. C., & Williams, C.

J. (1995). Variability in automatic activation as an

unobtrusivemeasure of racial attitudes: a bona fide

pipeline? Journal of Personality and Social

Psychology, 69, (6), 1013-1027.

Fazio, R.H., Powell, M.C., & Herr, P.M. (1983). Towards a

process model of the attitude-behavior relation:

Assessing one’s attitude upon mere observation of the

attitude object. Journal of Personality and Social

Psychology, 44, 723-735.

Fazio, R.H., Sanbonmatsu, D.M., Powell, M.C., & Kardes,

F.R. (1986). On the automatic activation of attitudes.

Journal of Personality and Social Psychology, 50, (2),

229-238.

90

Fazio, R. H. & Zanna, M. P. (1981). Direct experience and

attitude-behavior consistency. Advances in

Experimental Social Psychology, 14, 162-202.

Fazio, R. H., & Zanna, M. P. (1978). Attitudinal qualities

relating to the strength of the attitude-behavior

relationship. Journal of Experimental Social

Psychology, 1978, 14, 398-408.

Greenwald, A. G., Banaji, M. R., Rudman, L. A., Farnham, S.

D., Nosek, B. A., & Mellott, D. S. (2002). A unified

theory of implicit attitudes, stereotypes, self

esteem, and self-concept. Psychological Review, 109,

3-25.

Greenwald, A. G., & Banaji, M. R. (1995). Implicit social

cognition: Attitudes, self-esteem, and stereotypes.

Psychological Review, 102, 4-27.

Greenwald, A. G., & Farnham, S. D. (2000). Using the

implicit association test to measure self-esteem and

self-concept. Journal of Personality and Social

Psychology, 79, 1022-1038.

91

Greenwald, A. G., Klinger, M. R., & Liu, T. J. (1989).

Unconscious processing of dichoptically masked words.

Memory and Cognition, 17, 35-47.

Greenwald, A. G., McGhee, D. E., & Schwartz, J. K. L.

(1998). Measuring individual differences in implicit

cognition: The implicit association test. Journal of

Personality and Social Psychology, 74, 1464-1480.

Greenwald, A. G., Nosek, B. A., & Banaji, M. R. (2003).

Understanding and using the implicit association test:

I. An improved scoring algorithm. Journal of

Personality and Social Psychology, 85, 197-216.

Greenwald, A. G., & Nosek, B. A. (2001). Health of the

implicit association test at age 3. Zeitschrift für

Experimentelle Psychologie, 48, 85-93.

Hepner, P.P., Kivlighan, D.M., Wampold, B.E.(1992) Research

Design in Counseling. Pacific Grove, CA: Brooks/Cole.

Higgins, E.T., & King, G. (1981). Accessibility of social

constructs: Information processing consequences of

individual and context variability. In N. Cantor &

J.F. Kihlstrom (Eds.), Personality, cognition, and

social interaction. Hillsdale, N.J.: Erlbaum.

92

Hummert, M. L., Garstka, T. A., O'Brien, L. T., Greenwald,

A. G., Mellott, D. S. (2002). Using the implicit

association test to measure age differences in

implicit social cognitions. Psychology and Aging, 17,

482-495.

Jackson, D.N. (1967). Personality research form manual.

Goshen, NY: Research Psychologist Press, Inc.

Marlowe, D., & Crowne, D.P. (1960) A new scale of social

desirability independent of psychology. Journal of

Consulting Psychology, 24(4), 349-354.

Novaco, R.W. (1975) Anger control: The development and

evaluation of an experimental treatment. Lexington,

MA: D.C. Heath.

Perdue, C.W., Dovidio, J.F., Gurtman, M.B. & Tyler, R.B.

(1990). Us and them: Social categorization and the

process of intergroup bias. Journal of Personality and

Social Psychology, 59, 475-486.

Perdue, C.W., & Gurtman, M.B. (1990). Evidence for the

automaticity of ageism. Journal of Experimental

Social Psychology, 26, 199-216.

93

Rosenberg, M. (1965). Society and the adolescent self

image, Princeton University Press, Princeton,

N.J.Schofield, J.W. Effects of norms, public

disclosure, and need for approval on volunteering

behavior consistent with attitudes. Journal of

Personality and Social Psychology, 1975, 31, 1126

2233.

Sarbin, T.R. (1964). The rationality of nonsense: Intensity

of meaning of non-referential verbal units.

Psychological Record, 14(4), pp 401-410.

Schofield, J. W. (1975). Effects of norms, public

disclosure, and need of approval on volunteering

behavior consistent with attitudes. Journal of

personality and Social Psychology, 31, 1126-1133.

Spielberger, C.D. (1999). State-trait anger expression

inventory-2 (Professional Manual). Lutz, FL:

Psychological Assessment Resources.

Spielberger, C.D. (1996). State-trait anger expression

inventory-2 (professional manual). Lutz, FL:

Psychological Assessment Resources, Inc.

94

Spielberger, C.D. (1988). Manual for the state-trait anger

expression inventory. Odess, FL: Psychological

Assessment Resources.

Spielberger, C.D. (1983). State-trait anxiety inventory: A

comprehensive bibliography. Palo-Alto, CA: South FL.

Spielberger, C.D. (1966). Theory and research on anxiety.

In C.D. Spielberger (Ed.), Anxiety and behavior.

New York: Academic Press.

Spielberger, C.D., Gorsuch, R.L., & Lushene, R.E., Vagg

P.R., & Jacobs, G.A. (1983). State-trait anxiety

inventory (Form Y). Redwood City, CA: Mind Garden.

Spielberger, C.D., Gorsuch, R.L., & Lushene, R.E., Vagg

P.R., & Jacobs, G.A. (1970). State-trait anxiety

inventory (Self-Evaluation Questionaire). Palo Alto,

CA: Consulting Psychologist Press.

Spielberger, C.D., Jacobs, G., Russell, S., & Crane, R.S.

(1983). Assessment of anger: The state-trait anger

scale. Advances in Personality Assessment, 2, 161-189.

Spielberger, C.D., Vagg, P.R., Barker, L.R., Donham, G.W.,&

Westberry, L.G. (1980). The factor structure of the

95

State Trait Anxiety Inventory. In I.G. Sarason & C.D.

Spielberger (Eds.), Stress and anxiety (Vol. 7). New

York: Hemisphere/Wiley.

Spielberger, C.D., Vagg, P.R., Barker, L.R., & Westberry,

L.G. (1980). The factor structure of the state-trait

anxiety inventory. In I.G. Sarason & C.D.

Spielberger (Eds.), Stress and Anxiety (Vol. 7). New

York: Hemisphere/Wiley.

Strahan, R., & Gerbasi, K.C. (1972). Short, homogeneous

versions of the marlow-crowne social desirability

scale. Journal of Clinical Psychology, 28, 191-193.

Warner, L.G., & DeFleur, M.L. (1969). Attitude as an

interactional concept: Social constraint and social

distance as intervening variables between attitudes

and action. American Sociological Review, 34, 153-

169.

Zanna, M. P., Olson, J. M., & Fazio, R. H. (1980). Attitude

–behavior consistency: An individual difference

perspective. Journal of Personality and Social

Psychology, 38, 432-440.

96

Zuckerman, M. & Lubin, B. (1965). Manual for the multiple

affect adjective checklist. San Diego, CA: Educational

and Industrial Testing Service.

97

APPENDIX A

DEMOGRAPHIC DATA SHEET

1. Age ____

2. Gender

a. Male ____

b. Female ____

3. Ethnicity:

a. Caucasian ___

b. African-American ___

c. Hispanic ____

d. Asian ___

e. Native American ____

f. Other _____________

4. Grade Level __

5. Parent’s Occupation or Self Occupation ____________

6. Problems with anger:

a. problems with family ___

b. problems with the law ____

c. problems with a significant other ____

d. problems with strangers ____

98

APPENDIX B

99

100

101

APPENDIX C

102

103

APPENDIX D

A. Rosenberg Self-Esteem Scale

The scale is a ten-item Likert scale with items answered on a four-point scale from strongly agree to strongly disagree. The scoring for some items needs to be reversed so that in each case the scores go from less to more self-esteem. The original sample for which the scale was developed consisted of 5,024 High School Juniors and Seniors from 10 randomly selected schools in New York State.

Instructions: Below is a list of statements dealing with your general feelings about yourself. If you strongly agree, circle SA. If you agree with the statement, circle A. If you disagree, circle D. If you strongly disagree, circle SD.

Strongly Strongly Agree agree disagree disagree

1. On the whole, I am satisfied with myself.

2. *At times I think I am no good at all 3. I feel that I have a number of good qualities 4. I am able to do things as well as most other people 5. *I feel I do not have much to be proud of 6. *I certainly feel useless at times 7. I feel that I am a person on worth, at least on an equal plane with others 8. *I wish I could have more respect for myself 9. *All in all, I’m inclined to feel that I am a failure 10. I take a positive attitude toward myself

Note: Items with an asterisk are reversed scored.

104

APPENDIX E

Marlowe – Crowne Social Desirability Scale (short version)

Please answer true or false for each question.

1. I never hesitate to go out of my way to help someone in trouble. 2. I have never intensely disliked anyone.

3. There have been times when I was quite jealous of the good

fortune of others.

4. I would never think of letting someone else be punished for my

wrong doings.

5. I sometimes feel resentful when I don’t get my way.

6. There have been times when I felt like rebelling against people

in authority even though I knew they were right.

7. I am always courteous, even to people who are disagreeable.

8. When I don’t know something I don’t at all mind admitting it.

9. I can remember “playing sick” to get out of something.

10.I am sometimes irritated by people who ask favors of me.

105

APPENDIX F

Help out a fellow Aggie complete his dissertation

Contact

Rafael Cuellar at r-cuellar@neo.tamu.edu

Or (979)575-7290

You will asked to fill out forms that describe how you

feel and think.

One test is on a computer.

Location is at Harington Tower

Duration 1.5 hours

106

APPENDIX G

The Validation of the Implicit Association Anger Test Statement of Consent to Participate in Research

I understand that the purpose of this study is to

investigate the validity of an implicit association test. I was selected to be a possible participant because I am a student at Texas A&M University. A total of sixty people have been asked to participate in this study.

If I agree to be in this study I will be asked to complete self-reports written and computer form. This study will only take 1.5 hours and will take place in one sitting. The risks associated with this study are possible feeling uncomfortable if I do not like reporting my anxiety, anger, and self-esteem. The benefits of the participation include the $5.00 compensation, which will be awarded at the conclusion of the questionnaires.

This study is anonymous and I am aware that I will not put my name on the forms. I will be asked to make up a name that I place on all the forms. I understand that my name will not be given out to anyone and my forms will be in a locked university office. My decision whether or not to participate will not affect my current or future relations with Texas A&M University. If I decide to participate, I am free to refuse to answer any of the questions that may make me feel uncomfortable. I can withdraw at anytime without my relations to the university being affected. I can contact Rafael Cuellar, (210) 884-4517 (longhorn_ag@hotmail.com) or Dr. Collie Conoley, 862-3879 (collie-conoley@tamu.edu).

I understand that this research study has been reviewed and approved by the Institutional Review Board-Human Subjects in Research, Texas A&M University. For research-related problems or questions regarding subject’ rights, I can contact the Institutional Review Board through Dr. Michael W. Buckley, Director of Research Compliance, Office of the Vice President for Research at (979)845-8585 (mwbuckley@tamu.edu).

I have read and understand the explanation provided to me. I have had all my questions answered to my satisfaction, and I voluntarily agree to participate in this study. I have been given a copy of this consent form.

Signature of Subject:__________________ Date:____________ Signature of Investigator:_____________ Date:____________

107

APPENDIX H

Means and standard deviations were computed for the IAAT response error and latency, state anxiety, trait anxiety, social desirability scale, and self esteem scale. Subject Error Latency S-ANX T-ANX SDS SE

1 11.25 182.55 31 47 5 16 2 5 116.625 45 38 3 22 3 12.5 325.375 49 54 5 19 4 13.75 -386.925 47 54 1 21 5 1.25 -292.7 23 50 7 14 6 1.25 294.875 21 25 9 12 7 10 417.95 42 40 1 19 8 21.25 -7.65 24 37 3 11 9 20 134.4 51 60 7 29 10 12.5 415.4 49 51 6 23 11 12.5 949.9 24 47 3 26 12 20 -206.9 34 33 6 19 13 22.5 141.6 20 51 4 24 14 13.75 1039.25 33 37 6 21 15 11.25 352.775 27 42 6 21 16 33.75 60.8 32 47 6 21 17 28.75 -321.525 40 51 2 17 18 5 235.7 52 48 2 15 19 15 -366.425 25 27 2 14 20 13.75 113.825 39 48 5 15 21 3.75 79.275 34 37 2 21 22 10 -150.225 37 41 6 19 23 13.75 1068.125 26 35 5 12 24 20 -414.375 27 53 4 26 25 28.75 -1004.525 26 33 8 13 26 27.5 540.025 37 37 7 22 27 18.75 -84.2 48 48 6 13 28 8.75 446.225 30 34 8 14 29 5 -126.925 37 28 8 18 30 7.5 409.35 27 24 8 14

108

31 16.25 1075.325 20 20 10 10 32 23.75 608.775 22 26 9 10 33 20 -165.5 28 32 4 19 34 18.75 433.8 25 24 9 12 35 16.25 438.975 33 35 4 16 36 21.25 456.675 22 28 7 14 37 18.75 -94.425 38 40 7 11 38 8.75 407.425 51 43 7 17 39 13.75 275.475 33 31 4 19 40 11.25 430.9 21 29 5 13 41 7.5 -500.8 30 49 3 18 42 7.5 1120.675 40 45 5 19 43 3.75 -618.675 20 51 9 14 44 12.5 27.675 29 30 6 15 45 12.5 -371.65 32 33 4 15 46 1.25 -82 29 39 5 20 47 1.25 -118.675 43 29 4 21 48 1.25 281.625 32 29 9 12 49 0 109.075 48 46 7 20 50 18.75 -386.575 32 40 6 20 51 20 -212.85 52 60 4 24 52 8.75 17.075 20 30 8 20 53 8.75 -245.825 42 35 8 18 54 21.25 -240.075 30 27 6 20 55 35 367.775 54 46 3 23 56 16.25 102.8 50 30 4 17 57 1.25 133.775 31 32 6 13 58 2.5 -233.5 24 27 5 13 59 2.5 -306.05 20 20 7 15 60 0 65.175 33 29 8 20 Mean 13 112.30083 33.6833 38.2 5.56667 17.483STDEV 8.55192 427.10681 9.99415 10.084 2.19368 4.3185

109

APPENDIX I

Means and standard deviations were calculated for state anger and trait anger on the State-Trait Anger Expression Inventory-2.

S-Ang SAngF SAngV SAngP T-Ang TangT TangR 17 6 6 5 17 4 8 19 7 6 6 24 9 12 15 5 5 5 28 7 13 19 9 5 5 23 11 9 15 5 5 5 11 4 5 15 5 5 5 13 4 7 34 13 14 7 29 12 12 15 5 5 5 20 8 9 17 7 5 5 21 10 8 25 10 10 5 16 5 8 15 5 5 5 29 7 16 16 5 6 5 16 4 8 23 10 6 5 20 7 7 17 7 5 5 15 6 7 15 5 5 5 13 4 7 15 5 5 5 20 6 9 23 7 9 7 22 8 10 36 11 15 10 35 15 13 22 7 7 8 21 7 9 15 5 5 5 14 4 7 15 5 5 5 21 8 9 15 5 5 5 15 4 9 15 5 5 5 18 5 10 15 5 5 5 24 10 11 15 5 5 5 16 4 9 18 8 5 5 12 5 6 15 5 5 5 23 9 11 15 5 5 5 15 6 7 16 5 5 6 17 5 9 15 5 5 5 14 5 7

110

15 5 5 5 10 4 4 15 5 5 5 15 5 7 15 5 5 5 20 7 10 15 5 5 5 13 4 7 15 5 5 5 15 5 7 15 5 5 5 12 4 6 15 5 5 5 21 6 13 21 9 6 6 23 8 10 15 5 5 5 14 6 6 15 5 5 5 23 8 11 15 5 5 5 22 7 11 15 5 5 5 11 4 5 15 5 5 5 13 4 6 15 5 5 5 13 4 7 15 5 5 5 15 4 9 15 5 5 5 20 5 12 15 5 5 5 17 6 8 15 5 5 5 14 5 4 15 5 5 5 15 5 8 15 5 5 5 15 4 4 31 7 13 11 38 13 13 15 5 5 5 12 5 5 15 5 5 5 17 7 7 21 5 9 7 19 9 6 35 13 14 8 34 15 10 19 6 7 6 12 4 5 15 5 5 5 18 6 9 15 5 5 5 15 4 8 15 5 5 5 10 4 4 15 5 5 5 15 8 4 Mean 17.40 5.95 5.97 5.45 18.22 6.38 8.30 SD 5.05 1.94 2.38 1.18 6.07 2.71 2.65

111

APPENDIX J

Means and standard deviations were calculated for anger expression-in, anger expression-out, anger control-out, anger control-in, and anger expression index on the State-Trait Anger Expression Inventory-2.

AX-O AX-I AC-O AC-I AX

Index 18 15 22 26 33 18 25 17 18 56 16 25 23 24 62 16 16 8 8 64 10 9 20 11 36 10 8 32 32 2 24 21 17 16 60 17 17 16 20 46 19 18 17 16 52 13 21 31 20 31 19 27 22 21 51 18 13 30 32 17 11 13 15 19 38 14 14 29 26 21 12 15 25 17 33 15 17 16 16 48 20 17 18 13 54 26 28 22 22 58 19 18 17 13 54 15 12 28 23 24 18 20 26 25 35 12 20 28 20 32 14 14 30 25 21 16 21 16 10 59 13 14 32 27 16 23 18 14 16 59 14 15 23 25 29 21 21 26 27 37

112

12 13 24 22 27 17 13 30 22 26 11 26 32 32 21 12 10 30 26 14 14 13 25 21 29 10 17 29 26 20 16 13 20 22 35 10 11 32 31 6 14 19 28 21 32 16 21 17 20 48 15 13 23 23 30 18 12 26 30 22 16 15 20 19 40 9 18 26 19 30 13 11 31 31 10 8 8 32 32 0 12 14 28 31 15 14 18 21 17 42 14 14 27 22 27 11 15 29 27 18 14 20 24 23 35 12 14 28 31 15 26 21 18 17 58 13 17 30 28 20 10 15 30 26 17 17 14 21 20 38 19 20 25 16 46 8 13 16 16 40 16 21 22 26 37 15 14 28 21 28 8 8 32 32 0 14 11 23 19 31

Mean 14.92 16.23 24.12 22.28 33.08 SD 4.10 4.64 5.76 5.94 16.28

113

VITA

Rafael Cuellar, Jr. 1284 Farm Road 665 Alice, Texas 78332 Education B.A., Psychology, University of Texas, 1994 M.S., Counseling Psychology, Texas A&M University-Kingsville,

1995 Ph.D.,Counseling Psychology, Texas A&M University, 2005

Presentations

Morales, P.C., Bender, S.W., Hirsch, W.M., Howze, A.R., & Cuellar, R.(1998, March). The Uganda AIDS/HIV crisis:Positive living and dying in the face of an epidemic. Paper presented at the Association for Death Education and Counseling, Chicago, Illinois.

Morales, P.C., Howze, A.R., Cuellar, R., & Hirsch, W.M.

(1999, March).Quality of life in terminally ill individuals infected with HIV. Paper presented at the Association for Death Education and Counseling, San Antonio, Texas.